Suncelis in the Content Management System (CMS) Industry

Suncel

This paper is available on arxiv under CC 4.0 license. Authors: (1) Jeffrey Ouyang-Zhang, UT Austin (2) Daniel J. Diaz, UT Austin (3) Adam R. Klivans, UT Austin (4) Philipp Krähenbühl, UT Austin Table of Links Abstract & Introduction Related Works Preliminary Methodology Results Limitations Conclusion, Acknowledgements, & References Experimental Setting & Additional Results 7 Conclusion We present a method that efficiently scales thermodynamic stability prediction from single mutations to higher-order mutations. Our key insight is that the effects of mutations on the same protein are correlated. Thus, for a target protein, it suffices to run a deep backbone once and decode the effect of all mutations simultaneously using a shallow decoder. With the AlphaFold model as our backbone, our method outperforms existing methods on a variety of single and multiple mutation benchmarks. Our method scales to millions of mutations with minimal computational overhead and runs in a fraction of the time it would take prior works. 8 Acknowledgements This work is supported by the NSF AI Institute for Foundations of Machine Learning (IFML). References [1] Randomization of genes by pcr mutagenesis. Genome research 2(1), 28–33 (1992) 2 [2] Adams, J.P., Brown, M.J., Diaz-Rodriguez, A., Lloyd, R.C., Roiban, G.D.: Biocatalysis: A Pharma Perspective. Advanced Synthesis and Catalysis 361(11), 2421–2432 (2019). https://doi.org/10.1002/adsc.201900424 1 [3] Ahdritz, G., Bouatta, N., Kadyan, S., Xia, Q., Gerecke, W., O’Donnell, T.J., Berenberg, D., Fisk, I., Zanichelli, N., Zhang, B., Nowaczynski, A., Wang, B., Stepniewska-Dziubinska, M.M., Zhang, S., Ojewole, A., Guney, M.E., Biderman, S., Watkins, A.M., Ra, S., Lorenzo, P.R., Nivon, L., Weitzner, B., Ban, Y.E.A., Sorger, P.K., Mostaque, E., Zhang, Z., Bonneau, R., AlQuraishi, M.: Openfold: Retraining alphafold2 yields new insights into its learning mechanisms and capacity for generalization. bioRxiv (2022). https://doi.org/10.1101/2022.11.20.517210 6 [4] Arnold, F.H.: Design by directed evolution. Accounts of chemical research 31(3), 125–131 (1998) 1, 2, 3 [5] Bell, E.L., Finnigan, W., France, S.P., Green, A.P., Hayes, M.A., Hepworth, L.J., Lovelock, S.L., Niikura, H., Osuna, S., Romero, E., Ryan, K.S., Turner, N.J., Flitsch, S.L.: Biocatalysis. Nature Reviews Methods Primers 1(1), 1–21 (2021). https://doi.org/10.1038/s43586-021-00044-z 1 [6] Benevenuta, S., Pancotti, C., Fariselli, P., Birolo, G., Sanavia, T.: An antisymmetric neural network to predict free energy changes in protein variants. Journal of Physics D: Applied Physics 54(24), 245403 [7] Benevenuta, S., Pancotti, C., Fariselli, P., Birolo, G., Sanavia, T.: An antisymmetric neural network to predict free energy changes in protein variants. Journal of Physics D: Applied Physics 54(24), 245403 (2021) 8 [8] Benevenuta, S., Birolo, G., Sanavia, T., Capriotti, E., Fariselli, P.: Challenges in predicting stabilizing variations: An exploration. Frontiers in Molecular Biosciences 9, 1075570 (2023) 5, 6, 17 [9] Brückner, A., Polge, C., Lentze, N., Auerbach, D., Schlattner, U.: Yeast two-hybrid, a powerful tool for systems biology. International journal of molecular sciences 10(6), 2763–2788 (2009) 1 [10] Cadet, F., Saavedra, E., Syren, P.O., Gontero, B.: Machine learning, epistasis, and protein engineering: From sequence-structure-function relationships to regulation of metabolic pathways. Frontiers in Molecular Biosciences 9, 1098289 (2022) 1 [11] Capriotti, E., Fariselli, P., Casadio, R.: I-mutant2. 0: predicting stability changes upon mutation from the protein sequence or structure. Nucleic acids research 33(suppl_2), W306–W310 (2005) 3, 8 [12] Chen, K., Arnold, F.H.: Tuning the activity of an enzyme for unusual environments: sequential random mutagenesis of subtilisin e for catalysis in dimethylformamide. Proceedings of the National Academy of Sciences 90(12), 5618–5622 (1993) 1, 3 [13] Chen, T., Gong, C., Diaz, D.J., Chen, X., Wells, J.T., Wang, Z., Ellington, A., Dimakis, A., Klivans, A., et al.: Hotprotein: A novel framework for protein thermostability prediction and editing. In: The Eleventh International Conference on Learning Representations (2022) 3 [14] Chen, Y., Lu, H., Zhang, N., Zhu, Z., Wang, S., Li, M.: Premps: Predicting the impact of missense mutations on protein stability. PLoS computational biology 16(12), e1008543 (2020) 3, 8 [15] Cheng, J., Randall, A., Baldi, P.: Prediction of protein stability changes for single-site mutations using support vector machines. Proteins: Structure, Function, and Bioinformatics 62(4), 1125–1132 (2006) 3, 8 [16] Dehghanpoor, R., Ricks, E., Hursh, K., Gunderson, S., Farhoodi, R., Haspel, N., Hutchinson, B., Jagodzinski, F.: Predicting the effect of single and multiple mutations on protein structural stability. Molecules 23(2), 251 (2018) 4, 7 [17] Dehouck, Y., Kwasigroch, J.M., Gilis, D., Rooman, M.: Popmusic 2.1: a web server for the estimation of protein stability changes upon mutation and sequence optimality. BMC bioinformatics 12(1), 1–12 (2011) 3, 8 [18] Diaz, D.J., Gong, C., Ouyang-Zhang, J., Loy, J.M., Wells, J.T., Yang, D., Ellington, A.J., Dimakis, A., Klivans, A.R.: Stability oracle: A structure-based graph-transformer for identifying stabilizing mutations (2023). https://doi.org/10.1101/2023.05.15.540857 3, 4, 7, 8, 16, 17, 18, 19 [19] Diaz, D.J., Kulikova, A.V., Ellington, A.D., Wilke, C.O.: Using machine learning to predict the effects and consequences of mutations in proteins. Current Opinion in Structural Biology 78, 102518 (2023) 3 [20] d’Oelsnitz, S., Diaz, D.J., Acosta, D.J., Schechter, M.W., Minus, M.B., Howard, J.R., Do, H., Loy, J., Alper, H., Ellington, A.D.: Synthetic microbial sensing and biosynthesis of amaryllidaceae alkaloids. bioRxiv pp. 2023–04 (2023) 3 [21] d’Oelsnitz, S., Kim, W., Burkholder, N.T., Javanmardi, K., Thyer, R., Zhang, Y., Alper, H.S., Ellington, A.D.: Using fungible biosensors to evolve improved alkaloid biosyntheses. Nature Chemical Biology 18(9), 981–989 (2022) 1 [22] Ellefson, J.W., Gollihar, J., Shroff, R., Shivram, H., Iyer, V.R., Ellington, A.D.: Synthetic evolutionary origin of a proofreading reverse transcriptase. Science 352(6293), 1590–1593 (2016) 1 [23] Feng, X., Sanchis, J., Reetz, M.T., Rabitz, H.: Enhancing the efficiency of directed evolution in focused enzyme libraries by the adaptive substituent reordering algorithm. Chemistry–A European Journal 18(18), 5646–5654 (2012) 3 [24] Frazer, J., Notin, P., Dias, M., Gomez, A., Min, J.K., Brock, K., Gal, Y., Marks, D.S.: Disease variant prediction with deep generative models of evolutionary data. Nature 599(7883), 91–95 (2021) 9 [25] Gebauer, M., Skerra, A.: Engineered protein scaffolds as next-generation therapeutics. Annual Review of Pharmacology and Toxicology 60, 391–415 (2020). https://doi.org/10.1146/annurev-pharmtox-010818- 021118 1 [26] Gerasimavicius, L., Livesey, B.J., Marsh, J.A.: Correspondence between functional scores from deep mutational scans and predicted effects on protein stability. bioRxiv pp. 2023–02 (2023) 9 [27] Giver, L., Gershenson, A., Freskgard, P.O., Arnold, F.H.: Directed evolution of a thermostable esterase. Proceedings of the National Academy of Sciences 95(22), 12809–12813 (1998) 2, 3 [28] Hie, B.L., Shanker, V.R., Xu, D., Bruun, T.U., Weidenbacher, P.A., Tang, S., Wu, W., Pak, J.E., Kim, P.S.: Efficient evolution of human antibodies from general protein language models. Nature Biotechnology (2023) 3 [29] Jaroszewicz, W., Morcinek-Orłowska, J., Pierzynowska, K., Gaffke, L., W˛egrzyn, G.: Phage display and other peptide display technologies. FEMS Microbiology Reviews 46(2), fuab052 (2022) 1 [30] Järvelin, K., Kekäläinen, J.: Cumulated gain-based evaluation of ir techniques. ACM Transactions on Information Systems (TOIS) 20(4), 422–446 (2002) 6, 17 [31] Jumper, J., Evans, R., Pritzel, A., Green, T., Figurnov, M., Ronneberger, O., Tunyasuvunakool, K., Bates, R., Žídek, A., Potapenko, A., et al.: Highly accurate protein structure prediction with alphafold. Nature 596(7873), 583–589 (2021) 2, 3, 4, 6, 9, 10 [32] Kellogg, E.H., Leaver-Fay, A., Baker, D.: Role of conformational sampling in computing mutation-induced changes in protein structure and stability. Proteins: Structure, Function, and Bioinformatics 79(3), 830–838 (2011) 3 [33] Klabunde, T., Petrassi, H.M., Oza, V.B., Raman, P., Kelly, J.W., Sacchettini, J.C.: Rational design of potent human transthyretin amyloid disease inhibitors. Nature structural biology 7(4), 312–321 (2000) 1, 2 [34] Kouba, P., Kohout, P., Haddadi, F., Bushuiev, A., Samusevich, R., Sedlar, J., Damborsky, J., Pluskal, T., Sivic, J., Mazurenko, S.: Machine learning-guided protein engineering. ACS Catalysis 13, 13863–13895 (2023) 3 [35] Laimer, J., Hofer, H., Fritz, M., Wegenkittl, S., Lackner, P.: Maestro-multi agent stability prediction upon point mutations. BMC bioinformatics 16(1), 1–13 (2015) 3, 4, 7, 8 [36] Li, B., Yang, Y.T., Capra, J.A., Gerstein, M.B.: Predicting changes in protein thermodynamic stability upon point mutation with deep 3d convolutional neural networks. PLoS computational biology 16(11), e1008291 (2020) 3, 4, 8 [37] Li, G., Panday, S.K., Alexov, E.: Saafec-seq: a sequence-based method for predicting the effect of single point mutations on protein thermodynamic stability. International journal of molecular sciences 22(2), 606 (2021) 3 [38] Lin, Z., Akin, H., Rao, R., Hie, B., Zhu, Z., Lu, W., Smetanin, N., dos Santos Costa, A., Fazel-Zarandi, M., Sercu, T., Candido, S., et al.: Language models of protein sequences at the scale of evolution enable accurate structure prediction. bioRxiv (2022) 3, 4, 6, 9, 10, 18 [39] Loell, K., Nanda, V.: Marginal protein stability drives subcellular proteome isoelectric point. Proceedings of the National Academy of Sciences 115(46), 11778–11783 (2018) 1 [40] Lu, H., Diaz, D.J., Czarnecki, N.J., Zhu, C., Kim, W., Shroff, R., Acosta, D.J., Alexander, B.R., Cole, H.O., Zhang, Y., et al.: Machine learning-aided engineering of hydrolases for pet depolymerization. Nature 604(7907), 662–667 (2022) 3 [41] Meghwanshi, G.K., Kaur, N., Verma, S., Dabi, N.K., Vashishtha, A., Charan, P.D., Purohit, P., Bhandari, H.S., Bhojak, N., Kumar, R.: Enzymes for pharmaceutical and therapeutic applications. Biotechnology and Applied Biochemistry 67(4), 586–601 (2020). https://doi.org/10.1002/bab.1919 1 [42] Meier, J., Rao, R., Verkuil, R., Liu, J., Sercu, T., Rives, A.: Language models enable zero-shot prediction of the effects of mutations on protein function. Advances in Neural Information Processing Systems 34, 29287–29303 (2021) 9, 18 [43] Mirdita, M., Schütze, K., Moriwaki, Y., Heo, L., Ovchinnikov, S., Steinegger, M.: Colabfold: making protein folding accessible to all. Nature methods 19(6), 679–682 (2022) 6 [44] Montanucci, L., Capriotti, E., Frank, Y., Ben-Tal, N., Fariselli, P.: Ddgun: an untrained method for the prediction of protein stability changes upon single and multiple point variations. BMC bioinformatics 20, 1–10 (2019) 3, 4, 7, 8, 16, 18 [45] Montanucci, L., Savojardo, C., Martelli, P.L., Casadio, R., Fariselli, P.: On the biases in predictions of protein stability changes upon variations: the inps test case. Bioinformatics 35(14), 2525–2527 (2019) 16 [46] Musdal, Y., Govindarajan, S., Mannervik, B.: Exploring sequence-function space of a poplar glutathione transferase using designed information-rich gene variants. Protein Engineering, Design and Selection 30(8), 543–549 (2017) 3 [47] Nijkamp, E., Ruffolo, J., Weinstein, E.N., Naik, N., Madani, A.: Progen2: exploring the boundaries of protein language models. arXiv preprint arXiv:2206.13517 (2022) 9 [48] Nikam, R., Kulandaisamy, A., Harini, K., Sharma, D., Gromiha, M.M.: Prothermdb: thermodynamic database for proteins and mutants revisited after 15 years. Nucleic acids research 49(D1), D420–D424 (2021) 2, 16 [49] Notin, P., Dias, M., Frazer, J., Hurtado, J.M., Gomez, A.N., Marks, D., Gal, Y.: Tranception: protein fitness prediction with autoregressive transformers and inference-time retrieval. In: International Conference on Machine Learning. pp. 16990–17017. PMLR (2022) 2, 3, 9, 16, 18 [50] Notin, P.M., Van Niekerk, L., Kollasch, A.W., Ritter, D., Gal, Y., Marks, D.: Trancepteve: Combining family-specific and family-agnostic models of protein sequences for improved fitness prediction. bioRxiv pp. 2022–12 (2022) 9 [51] Nutschel, C., Fulton, A., Zimmermann, O., Schwaneberg, U., Jaeger, K.E., Gohlke, H.: Systematically scrutinizing the impact of substitution sites on thermostability and detergent tolerance for bacillus subtilis lipase a. Journal of chemical information and modeling 60(3), 1568–1584 (2020) 9 [52] Paik, I., Ngo, P.H., Shroff, R., Diaz, D.J., Maranhao, A.C., Walker, D.J., Bhadra, S., Ellington, A.D.: Improved bst dna polymerase variants derived via a machine learning approach. Biochemistry (2021) 3 [53] Pak, M.A., Markhieva, K.A., Novikova, M.S., Petrov, D.S., Vorobyev, I.S., Maksimova, E.S., Kondrashov, F.A., Ivankov, D.N.: Using alphafold to predict the impact of single mutations on protein stability and function. Plos one 18(3), e0282689 (2023) 3 [54] Pancotti, C., Benevenuta, S., Birolo, G., Alberini, V., Repetto, V., Sanavia, T., Capriotti, E., Fariselli, P.: Predicting protein stability changes upon single-point mutation: a thorough comparison of the available tools on a new dataset. Briefings in Bioinformatics 23(2), bbab555 (2022) 2, 8, 16, 17 [55] Pancotti, C., Benevenuta, S., Repetto, V., Birolo, G., Capriotti, E., Sanavia, T., Fariselli, P.: A deep-learning sequence-based method to predict protein stability changes upon genetic variations. Genes 12(6), 911 (2021) 8 [56] Pires, D.E., Ascher, D.B., Blundell, T.L.: Duet: a server for predicting effects of mutations on protein stability using an integrated computational approach. Nucleic acids research 42(W1), W314–W319 (2014) 3, 8 [57] Pires, D.E., Ascher, D.B., Blundell, T.L.: mcsm: predicting the effects of mutations in proteins using graph-based signatures. Bioinformatics 30(3), 335–342 (2014) 8 [58] Pucci, F., Schwersensky, M., Rooman, M.: Artificial intelligence challenges for predicting the impact of mutations on protein stability. Current opinion in structural biology 72, 161–168 (2022) 17 [59] Qiu, Y., Wei, G.W.: Persistent spectral theory-guided protein engineering. Nature Computational Science pp. 1–15 (2023) 3, 6, 17 [60] Rao, R.M., Liu, J., Verkuil, R., Meier, J., Canny, J., Abbeel, P., Sercu, T., Rives, A.: Msa transformer. In: International Conference on Machine Learning. pp. 8844–8856. PMLR (2021) 3, 9, 10 [61] Riesselman, A.J., Ingraham, J.B., Marks, D.S.: Deep generative models of genetic variation capture the effects of mutations. Nature methods 15(10), 816–822 (2018) 9 [62] Rives, A., Meier, J., Sercu, T., Goyal, S., Lin, Z., Liu, J., Guo, D., Ott, M., Zitnick, C.L., Ma, J., Fergus, R.: Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences. PNAS (2019). https://doi.org/10.1101/622803 3, 7 [63] Rodrigues, C.H., Pires, D.E., Ascher, D.B.: Dynamut: predicting the impact of mutations on protein conformation, flexibility and stability. Nucleic acids research 46(W1), W350–W355 (2018) 3, 8 [64] Rodrigues, C.H., Pires, D.E., Ascher, D.B.: Dynamut2: Assessing changes in stability and flexibility upon single and multiple point missense mutations. Protein Science 30(1), 60–69 (2021) 3, 4, 7 [65] Sanavia, T., Birolo, G., Montanucci, L., Turina, P., Capriotti, E., Fariselli, P.: Limitations and challenges in protein stability prediction upon genome variations: towards future applications in precision medicine. Computational and Structural Biotechnology Journal 18, 1968–1979 (2020). https://doi.org/10.1016/j.csbj.2020.07.011, https://doi.org/10.1016/j.csbj.2020.07.011 6, 17 [66] Savojardo, C., Fariselli, P., Martelli, P.L., Casadio, R.: Inps-md: a web server to predict stability of protein variants from sequence and structure. Bioinformatics 32(16), 2542–2544 (2016) 3, 8 [67] Schymkowitz, J., Borg, J., Stricher, F., Nys, R., Rousseau, F., Serrano, L.: The foldx web server: an online force field. Nucleic acids research 33(suppl_2), W382–W388 (2005) 3, 7, 8, 18 [68] Shroff, R., Cole, A.W., Diaz, D.J., Morrow, B.R., Donnell, I., Annapareddy, A., Gollihar, J., Ellington, A.D., Thyer, R.: Discovery of novel gain-of-function mutations guided by structure-based deep learning. ACS synthetic biology 9(11), 2927–2935 (2020) 3 [69] Starr, T.N., Thornton, J.W.: Epistasis in protein evolution. Protein science 25(7), 1204–1218 (2016) 1 [70] Steinegger, M., Soding, J.: Mmseqs2 enables sensitive protein sequence searching for the analysis of massive data sets. Nature biotechnology 35(11), 1026–1028 (2017) 6 [71] Stemmer, W.P.: Rapid evolution of a protein in vitro by dna shuffling. Nature 370(6488), 389–391 (1994) 2, 3 [72] Tsuboyama, K., Dauparas, J., Chen, J., Laine, E., Mohseni Behbahani, Y., Weinstein, J.J., Mangan, N.M., Ovchinnikov, S., Rocklin, G.J.: Mega-scale experimental analysis of protein folding stability in biology and protein design. bioRxiv pp. 2022–12 (2022) 1, 2, 6, 16 [73] Tung, J.W., Heydari, K., Tirouvanziam, R., Sahaf, B., Parks, D.R., Herzenberg, L.A., Herzenberg, L.A.: Modern flow cytometry: a practical approach. Clinics in laboratory medicine 27(3), 453–468 (2007) 1 [74] Umerenkov, D., Shashkova, T.I., Strashnov, P.V., Nikolaev, F., Sindeeva, M., Ivanisenko, N.V., Kardymon, O.L.: Prostata: Protein stability assessment using transformers. bioRxiv pp. 2022–12 (2022) 3, 4, 6, 7, 8, 18, 19 [75] Wittmann, B.J., Johnston, K.E., Wu, Z., Arnold, F.H.: Advances in machine learning for directed evolution. Current opinion in structural biology 69, 11–18 (2021) 3 [76] Wittmann, B.J., Yue, Y., Arnold, F.H.: Informed training set design enables efficient machine learningassisted directed protein evolution. Cell Systems 12(11), 1026–1045 (2021) 3 [77] Worth, C.L., Preissner, R., Blundell, T.L.: Sdm—a server for predicting effects of mutations on protein stability and malfunction. Nucleic acids research 39(suppl_2), W215–W222 (2011) 3, 8 [78] Wu, S., Snajdrova, R., Moore, J.C., Baldenius, K., Bornscheuer, U.T.: Biocatalysis: Enzymatic Synthesis for Industrial Applications. Angewandte Chemie - International Edition 60(1), 88–119 (2021). https://doi.org/10.1002/anie.202006648 1 [79] Yeung, N., Lin, Y.W., Gao, Y.G., Zhao, X., Russell, B.S., Lei, L., Miner, K.D., Robinson, H., Lu, Y.: Rational design of a structural and functional nitric oxide reductase. Nature 462(7276), 1079–1082 (2009) 1, 2 [80] Zhou, B., Lv, O., Yi, K., Xiong, X., Tan, P., Hong, L., Wang, Y.G.: Accurate and definite mutational effect prediction with lightweight equivariant graph neural networks. arXiv preprint arXiv:2304.08299 (2023) 3 [81] Zhou, Y., Pan, Q., Pires, D.E., Rodrigues, C.H., Ascher, D.B.: Ddmut: predicting effects of mutations on protein stability using deep learning. Nucleic Acids Research p. gkad472 (2023) 3

Predicting a Protein’s Stability
under a Million Mutations: Conclusion, Acknowledgement & References

When you work remotely, having the freedom to travel is a big part of the appeal for many. Colombia is a popular destination on the circuit and the recent introduction of a is likely to boost its appeal further. But sometimes, connectivity and internet access can be a problem, especially in locations that are slightly off the beaten track. digital nomad visa Let’s explore why digital nomads are drawn to Colombia, what the digital nomad visa offers and which locations offer the most reliable internet connections. Why digital nomads are drawn to Colombia Colombia is known as one of the most biodiverse countries in the world, with incredible landscapes, such as the Sierra Nevada and Andes mountain ranges. This country is also home to several iconic nature reserves including Tayrona National Park and Chiribiquete National Park, considered World Heritage by Unesco. And the country also offers plenty of cultural attractions too. Barranquilla, a city on the northern coast, hosts the second largest carnival in the world (after Rio de Janiero, Brazil) and travelers can also enjoy a number of festivals throughout the year, including “Feria de Cali” and “Feria de las Flores” in Medellín. Colombia Cost of Living The cost of living is typically low compared to other countries in South America, the official currency is the Colombian Peso (COP), and for an individual is about USD $800, adding to the attraction for many international workers. the average monthly living costs So, it comes as no surprise that Colombia is a big tourist destination, attracting in 2022. Most travelers enter on a tourist visa that offers three months on arrival with the option to extend your stay for a further three months once in the country. However, this visa is meant for leisure purposes and doesn’t properly support those wishing to work remotely in the country. While other visas are available, they’re often time consuming and complicated to obtain. 3.5 million tourists For this reason Colombia has been issuing new regulations in order to improve communication and make the application process more welcoming to remote workers and digital nomads who wish to stay in the country for a longer period. The Launch of the Digital Nomad Visa After a long legislative process, . The new visa will allow foreigners to live in the country for 24 months and access a state ID known as “Cedula de Extranjeria” which is needed to access many services in the country. Without this ID it can be difficult to obtain a cellular phone service contract or rent an apartment to name a couple of examples. Colombia launched its Digital Nomad Visa in 2023 The proposal was presented by Ilana Milkes, the CEO of Tech World, at the last Mobile World Congress, after two years of working on it alongside the government agency, . This represents a massive step for the country, considering the growing tech scene in major cities like Bogotá and Medellín and digital nomads’ preferences of having Colombia as one of their moving options. Procolombia The initiative is expected to facilitate the arrival of at least 45,000 digital nomads to the country in the first 18 months. However, considering the visa targets remote workers who need optimal internet speeds, how does Colombia fare when it comes to reliable connectivity? Colombia’s connectivity: Stats and facts According to several surveys, the South American region has been expanding its internet penetration, . Additionally, Colombia is currently in the fourth position of fastest growing internet right after Argentina, Mexico, and Brazil. coming in fifth behind North America Colombia now counts over 39.34 million internet users, and its internet penetration was at . The country is also improving its Network Readiness Index faster than the world average due to from Colombia’s Ministry of Information and Communication Technologies (MinTic). 75.7% at the beginning of 2023 initiatives So, while Colombia’s connectivity rates are slightly behind some of its close neighbors in Latin America, its government’s various development initiatives to improve internet connectivity may be paying off with internet penetration increasing steadily for the past three years. Regional internet providers and coverage across Colombia The country has a fixed broadband speed offering a minimum of 25 Mbps downstream and 5 Mbps upstream. In addition, internet providers offer fiber-optic internet as an option and mobile broadband speeds of up to 4G. As for the internet provider options available for digital nomads in Colombia, the four, main providers offering the following speeds, according to : speedtest.net Movistar with the fastest median download speed at 141.32 Mbps ETB with a download speed of 74.47Mbps Claro with a download speed of 88.86Mbps Tigo with a download speed of 75.88 While these speeds are the average, they do vary from region to region across Colombia. Let’s take a closer look at how some of the most popular cities across the country fare with their . fixed broadband speeds Larger cities: Bogotá: 51.92 Mbps Medellín: 46.9 Mbps Cali: 51.18 Mbps Bucaramanga: 52.81 Mbps Smaller cities: Santa Marta: 36.77 Mbps Villavicencio: 42.52 Mbps Sincelejo: 33.53 Mbps Montería: 32.79 Mbps With this data it’s clear that Colombia offers many reliable places to work from for the growing number of digital nomads. But what about those who want to get away from the cities? Working reliably in rural Colombia As mentioned, the country has a lot to offer in terms of places to visit; beaches, jungles, mountains and vast, lush plains are some of Colombia’s biggest attractions. Major Colombian cities have a lot to offer, but many digital nomads often begin to look outside the city after a while of living in the country. Here, connectivity becomes even more important to consider. If you’re hoping to work remotely from one of Colombia’s many natural beauty spots it’s important to be aware that Wifi coverage may not be as reliable so you can plan accordingly. For example, if you head just outside of the cities of Medellín or Santa Marta to little mountain getaways like Guatapé or Minca, then connectivity can quickly become harder to guarantee, despite their close proximity to the cities. Equally, there are many rural locations that are popular with nomads such as Salento, Salamina, Villa de Leyva, and Valle del Cocora that are likely to have slower internet speeds. While Wifi is standard in hotels or hostels, it’s advisable to have the back-up of a local network provider if you’re planning to work remotely in rural Colombia. Movistar may offer the fastest speeds in urban areas but its coverage in rural areas is limited. If you’re planning to use your mobile network to support your connectivity, then Claro is a safer bet, offering coverage across more rural locations. The rise of fiber optic and 5G across Colombia Looking ahead, Colombia already has the most extensive fiber optic network in South America, and is on its way to offer fiber optic broadband to by the end of 2023. 4 million homes and businesses As for the future of mobile services, Colombia is still in the process of . Once in place, 5G coverage will drastically improve internet coverage in rural areas and is noted as a going forward. adopting 5G 5G mobile networks top priority for MinTec Overall, the launch of Colombia’s new digital nomad visa is likely to attract many more remote workers to the country. While big cities offer internationally competitive internet speeds, these rates aren’t guaranteed across the entire country. Yet with a little forward planning and knowledge on Colombia’s network coverage it’s possible to work remotely without issue. This article was originally published by Stefania Troncoso on The Sociable.

Colombia's Digital Nomad Visa Raises Internet Connectivity Concerns

কলম্বিয়ার ডিজিটাল যাযাবর ভিসা ইন্টারনেট সংযোগের উদ্বেগ উত্থাপন করেছে৷

Kolumbiens digitales Nomadenvisum wirft Bedenken hinsichtlich der Internetverbindung auf

La visa de nómada digital de Colombia genera preocupaciones sobre la conectividad a Internet

Le visa nomade numérique colombien soulève des problèmes de connectivité Internet

कोलंबिया के डिजिटल खानाबदोश वीज़ा ने इंटरनेट कनेक्टिविटी की चिंता बढ़ा दी है

コロンビアのデジタル遊牧民ビザがインターネット接続の懸念を提起

콜롬비아의 디지털 유목민 비자로 인해 인터넷 연결 문제가 제기됨

Visto de nômade digital da Colômbia levanta preocupações com a conectividade da Internet

Kolombiya'nın Dijital Göçebe Vizesi İnternet Bağlantısı Kaygılarını Artırıyor

Thị thực du mục kỹ thuật số của Colombia làm tăng mối quan tâm về kết nối Internet

哥伦比亚的数字游牧签证引发互联网连接问题

For many years one of the main means of cancer diagnosis was a biopsy. A doctor would take a sample of the tissue in question, and a pathologist, among other tests, would make a thorough examination of it under the microscope. They would look for the affected cells, or for some other signs of developing the disease. That being said, this process is and was quite tedious, as the abnormal cells are minuscule compared to the enormous image of the tissue – the actual whole slide image. Modern histopathology involves multiple methods to analyze human cells, including tissue scans and anamnesis data. Today, healthcare organizations rely on or to save a digital scan of the entire histology result. WSI scans help doctors to analyze larger areas to detect malignant cells or other mutations in specific organ tissue. Whole Slide Imaging WSI However, the process of working with WSI is enormously time-consuming, since such slides usually have large resolutions. Running through them manually is a tough and time-consuming task. Given that this problem exists, the MobiDev team challenged themselves to search for an AI pathology solution capable of working with WSI and supporting digital healthcare workflows. Image credits to National Cancer Institute. What is WSI? WSI image is a high-resolution, large format scan of human tissue. WSI images are also called microscope slides, as they represent an enlarged image of cells and subcellular structure. Such slides are obtained by making multiple scans of different areas of a tissue sample and montaging them into one large image. As an output, we receive a high-quality digital scan that can be zoomed in up to 40 times without losing much detail. Compared to the traditional microscopic analysis of histology, WSI provides a number of advantages: Whole slide images usually come in giant resolutions like 100,000 x 100,000 pixels and bigger, which gives a lot of clarity when visually inspecting tissue areas. WSI is a digital scan that covers much larger areas all at once and is easier to zoom in/zoom out, compared to a microscopic view. The image can be sent to the healthcare experts as a digital copy. Additionally, WSI scans can’t deteriorate like specimens on a glass slide. However, in terms of operability, WSI analysis is not so great due to the time and effort needed during the manual examination. In real-world practice, a doctor would sit before the image scanning by eye, zooming in and out for hours. In the case of manual examination, a whole slide should be examined to spot suspicious cell structures or mutations, because skipping some areas may result in medical error. This is a general problem for healthcare workflows since long processes prevent quick diagnosis and efficient treatment, which are critical for patients with cancer and for histopathology as a whole. With the advances in digital pathology and fluctuations in healthcare regulatory requirements, the application of artificial intelligence became possible to use to process clinical data. Given the fact that WSI is a gigapixel-sized image document that contains tons of unstructured data, the problem with manual examination can be addressed through machine learning. Applying AI for WSI scans processing Independent of the content in the image, WSI allows for the image to be broken down into features which are used for training machine learning models that are capable of running object detection, recognition, and classification tasks. These tasks are handled with computer vision that relies on optical sensors or digitally processes images to extract meaningful information. However, since we don’t know where the potential tumor cells are, we can’t perform one essential ML task: data labelling. Labels are required to tell the model what these or those cells are for it to learn the difference between normal and damaged tissue. To solve this, convolutional neural networks, or CNNs, are used. CNN requires little pre-processing of data input, as they are mainly used for visual data to find patterns autonomously. Applied to WSI scans, CNN is capable of extracting features out of weakly labelled, or completely unlabeled data. Extracted features then can be used to represent the whole slide in compact form, which can be helpful for image storing or grouping similar items. However, minor pre-processing steps are still required, since it’s too CPU-demanding for the model to work with the whole slide. Due to the giant image size, the image is most likely to be sliced into equally-sized square images. So now, let’s look at the example in practice. MobiDev’s Research: WSI processing through CLAM approach With their extensive expertise in computer vision, Mobidev approached the WSI processing system design through clustering. This is a constrained attention multiple instance learning ( ) approach. As a training dataset, publicly available Clear Cell Renal Cell Carcinoma was used. CLAM dataset To prepare data for training, images need to be cut into patches (equally sized square images), because the initial image size is too large. Also, the data used for the experiment was weakly labeled, meaning there were no labels for each patch of images. The task was to create a model capable of analyzing WSI scans and then highlight regions of interest, or in other words, suspicious cells, found in the tissue. As long as healthcare is based on critical decisions, the role of AI in pathology is to support the doctor with hints to speed up the diagnostic process. However, the prediction should also be descriptive, as long as the user has to understand how the model derived this or that result. This is what’s called AI explainability as a feature of modern AI pathology systems. You can read more about the forms and approaches in AI explainability in decision-making applications. CLAM is a data-efficient approach to working with weakly-labeled whole-slide imagery. The essence of it can be broken down into a number of steps: On the WSI, the tissue/cell sample is segmented from the background using computer vision algorithms. Within the segmented area, the tissue is split in equally-sized square patches (usually 256px). Patches are encoded once by a pretrained CNN into a descriptive feature representation. During training and inference, extracted patches in each WSI are passed to a CLAM model as feature vectors. For each class, the attention network ranks each region in the slide and assigns an attention score based on its relative importance to the slide-level diagnosis. Attention-pooling weighs patches by their respective attention scores and summarizes patch-level features into slide-level representations, which are used to make the final diagnostic prediction. The attention scores can be visualized as a heatmap to identify ROIs and interpret the important features. Left: direct visualization of the attention scores. Each pixel represents a 256px patch. Right: attention scores blended onto the WSI. The colored regions represent the most attended parts of the sample. As we can see, the CLAM’s attention network is quite interpretable. It outputs “importance scores” for each patch, so those can be visualized as a heatmap. If you take a closer look at the WSI on the right, you may see the affected cells colored blue, meaning the network determined they would affect the prediction the most. Using neural networks to process weakly labeled WSI images provides a couple of advantages in this case. First of all, we can discover clinically relevant, but previously unrecognized morphological characteristics that have not been used in visual assessment by pathologists. Secondly, the model can recognize features that are beyond human perception, which can potentially provide more accurate diagnostic results and have a positive impact for surgical or therapeutic results. Addressing the WSI complexity and interpretability This approach addresses important WSI processing difficulties. For example, to handle the large size of the slide, we segment the actual tissue from the scan background, trimming unnecessary information. Then, we split the remaining ROI into tiles that are treated equally to fit them into the GPU. Also, another difficulty is labeling. It takes hours for a pathologist to look through the slide and diagnose, let alone label it, i.e. identifying affected cells. On the contrary, CLAM is designed to work with this kind of weakly-labeled data, where only the slide-level labels are known (i.e. diagnosis.) CLAM itself determines which regions of the tissue affect the predicted diagnosis. Written by , AI Engineer at . Anton Kantsemal MobiDev The full article was originally published and is based on MobiDev technology research. here

How AI Is Helping To Diagnose Cancer

We want our software to be error-proof, but in reality, error scenarios will always exist. So this article's objective is to explain why you should handle errors on your app, give you a rule of thumb on when treating errors, and give some practical improvements you could apply on your app. So, why handling errors? Finding errors asap Worse than finding a bug on your production app is learning that the bug you just found is around for several releases. Below is the recipe of failure to ship a hard to catch bug. { phone = telefoneNumber { } } func callMerchant (with telefoneNumber: String?) guard let else return // Calling a phone number code goes here If for some reason this function is ever invoked with a , it will exit without trying to make the phone call, and this could lead to a useless call button on your user interface, that does nothing when pressed. nil telefoneNumber Improving User Experience (UX) It might be really frightening to a user with no technology background to be prompted with an error dialog full of tech words and error codes, especially if he just made a critical action like a purchase. Take a look at the picture on the left. On the other hand, the picture to the right explains what happened and how the user could proceed. Recovering from an error state to a success With a really well-crafted error recovering strategy you could even recover a user that got in an error state, to the main flow of your application that will lead to a goal (like making a purchase). Recovering from errors is not only important for the tech team, but it is also beneficial for the business as a whole. Commonly, digital products lose some conversion percentual points due to techinical issues. And good error handling might mitigate this issue. The example above gives clear instructions and even some shortcuts on how to get out of this error and try another product. Rule of thumb to handling errors Maybe your App is nothing like any of the examples I gave so far, and you are not sure where you could improve error handling in your app. So you could follow this rule of thumb to know where you should consider handling errors. Consider handling errors every time you... ...make a request to an external source (networking) ...capture user input ...encode or decode some data ...escape a function prior to its full execution (early return) Practical improvements for your App Monitoring tool This is the most important improvement that you could do! With a monitoring tool, you can have useful data to discover, understand, and prioritize errors. By understanding the volumetry of an error, you could decide between adding a fix to the next version, creating a new version as soon as possible just to fix that error (hotfix), or turn off some remote configuration to disable the problematic feature. There are several monitoring tools available on the market, like or . The monitoring tool that I use at iFood is . It provides all the utility that we need to keep track of error logs: Dynatrace New Relic Logz.io Logs over time Querying for specific logs Configuring alerts to send to Slack Dashboard creation With a good tool setup in the project, it is time to bring a monitoring culture to the team. You could start establishing some new tasks that should be done at every new feature development. Map all error cases Create logs for the error cases Create alerts for the logs to get any critical scenario. It is important that the alerts are sent to a channel where all the devs have access. Create a Dashboard containing all the logs for that feature Monitor the dashboard periodically. You could make a recurrent event on the calendar to be reminded. Swift's Error protocol Swift's error protocol allows you to create expressive errors that will give you useful information to find and act on a possible issue. Having rich errors will also help you to create insightful dashboards and precise alerts on your monitoring tool. Below, there is a simple example of how to use it in an enumerator. { generic network(payload: [ : ])
} : enum SimpleError Error case case String Any In this example, we will capture the payload of the request that caused a network error. Doing so, we could look for patterns on the payloads, and understand what really causes the error. : if you want to do an error treatment like that one where you send the payload from a network request, you must mask any user sensitive data. Disclaimer You could also conform to this protocol in a so you can have as much information as you need on the error. This is useful when you want to custom tailor an error for a very specific scenario. struct { { one two
    } line: file: type: isUserLoggedIn: } { (line: , file: , type: .one, isUserLoggedIn: )
} : struct StructError Error enum ErrorType case case let Int let String let ErrorType let Bool // ... func functionThatThrowsError throws throw StructError 53 "main.swift" false Later on your monitoring tool, you could create a dashboard comparing the volumetry when the user has logged in again when it is not logged in, so you can understand if this is a relevant factor that leads to the error. And if you want the error to also contain localized human-readable messages to display, you could also conform to as shown below. LocalizedError { emptyName invalidEmail invalidPassword
} { errorDescription: ? { { .emptyName: .invalidEmail: .invalidPassword: }
    }
} : enum RegisterUserError Error case case case : extension RegisterUserError LocalizedError // errorDescription is the one that you get when using error.localizedDescription var String switch self case return "Name can't be empty" case return "Invalid email format" case return "The password must be at least 8 characters long" Now if an error of type happens, you could display to the user. RegisterUserError error.localizedDescription Don't use as an error nil Take a look at the function below, it is uncanny how familiar this code is. -> ? { dataFromKey = .standard.data.(forKey: ) data = dataFromKey { } decoder = () userPreferences = ? decoder.decode( . , from: data) userPreferences
} func getUserPreferences () UserPreferences let UserDefaults "user_preferences" guard let else return nil let JSONDecoder let try UserPreferences self return At first glance, pretty standard implementation. Nothing wrong with that. But if this code ever returns , how would you know if there are no set yet or if there is something wrong with our encoding or decoding of this object? nil UserPreferences If we want to ship this code to production it would be really nice that the caller of this function would create and save a new with default preferences if there are none set, or log an error to our monitoring tool if the decoding failed, so we could investigate and fix it. UserPreferences Returning when some error occurs really limits the options you have to handle it. nil You could instead make the function throwable, declaring it with and removing the optional mark from the decoder's . If you never used in Swift, I strongly recommend you read this , as I'll not cover the basics on what it is and how it works. throws ? try throws article from Sundell -> { dataFromKey = .standard.data.(forKey: ) data = dataFromKey { .noUserPreferences
    } decoder = () userPreferences = decoder.decode( . , from: data) userPreferences
} func getUserPreferences () throws UserPreferences let UserDefaults "user_preferences" guard let else throw UserPreferencesError let JSONDecoder let try UserPreferences self return on the example above, it would be correct to return on the clause of the , as is in fact the representation of the absence of value. I went all-in on throws just to illustrate better how it could be used. Note: nil else guard nil But what if the errors happen in an asynchronous context? Is it time to return ? Still, not ideal, instead, we could use other Swift's nice feature for error handling that suits really well in asynchronous contexts: . nil Result is an with the form that has a case with the result of the request as an associated value, and a case that brings an associated. Again, if you never heard of before, I strongly recommend this quick read from . Result enum Result<Success, Failure> where Failure: Error success failure Error Result hacking with swift If our fetch its data from a server instead of , we could rewrite it like the example below. getUserPreferences UserDefaults -> ) { .request(.userPreferences(userID: id)) { result result { .success( data): { decoder = () userPreferences = decoder.decode( . , from: data)
              completion(.sucess(userPreferences))
          } {
              completion(.failure(error))
          } .failure( error):
          completion(.failure(error))
      }
    }
} func getUserPreferences (userID id: String, completion: @escaping (Result<UserPreferences, Error>) Void Network in switch case let do let JSONDecoder let try UserPreferences self catch case let This way, the caller function could differentiate an encoding error from a network error, and log it. Separate error handling from the actual functionality If you are familiar with the principles, you know the importance of the . The says that our software units should have a single responsibility. What is a responsibility depends on the size of the software unit, the responsibility that a function can have is more narrow than the responsibility that a class or a module could handle. SOLID Single Responsibility Principle (SRP) SRP Take a look on the example below: { user.name.isEmpty == { .emptyName
    } isValid(email: user.email) { .invalidEmail
    } isValid(password: user.password) { .invalidPassword
    } } func registerUser ( user: User) _ throws guard false else throw RegisterUserError guard else throw RegisterUserError guard else throw RegisterUserError /*
        Code that registers a user goes here
    */ The function breaks and it is not easy to read because the code that actually register the user is at the end of the function, with all the validation rules first. registerUser SRP We could greatly improve this function by separating the error handling responsibility from the user registering responsibility. { validateUser(user) } { user.name.isEmpty == { .emptyName
    } isValid(email: user.email) { .invalidEmail
    } isValid(password: user.password) { .invalidPassword
    }
} func registerUser ( user: User) _ throws try /*
        Code that registers a user goes here
    */ func validateUser ( user: User) _ throws guard false else throw RegisterUserError guard else throw RegisterUserError guard else throw RegisterUserError Conforming to the made this function that much easier to read. SRP Recap 😌 Don't leave errors unhandled, your users will appreciate it. ☁️ Use a monitoring tool on your project. ❤️ Use Swift's protocol to get expressive and useful errors. Error 🙅‍♂️ Don't use as an error. nil 👨‍💻 Separate error validation and treatment from the actual functionality. What's next? If you are interested in error handling, chapter 6 of is a must-read. This chapter was the biggest reference to this article, and it was what really drew my attention to the importance of handling errors. I consider this book a must-have for any software engineer, so you might as well . Clean Code from Robert C. Martin (a.k.a Uncle Bob) get it on Amazon If you want to know more about and the other principles, is a great light start. And gives nice and easy examples and has links to in-depth articles that are really good reading. SRP SOLID this article at hackernoon This repo from ochococo Thank you for reading! I hope this is was insightful and that you can apply those ideas on your projects. Take care and good error handling! Previously published at https://lucasoliveira.tech/posts/improving-error-handling-in-your-app

Effective Error Handling in iOS

Andy: “Okay people, we need to hire some new talent for a big client. I’m going to need someone who can get at least 40 questions right on a 50-question multiple-choice test in 15 minutes. Probably have to guess some of the answers because of the time constraint. Do we have someone who can do that?” Right Hand Man: “Absolutely. Rob has all the test-prep files. I’ll ask him to start memorizing.” Andy: “Great. Now I need someone who got good grades in school and can write 50 high-performing topic sentences in a Google spreadsheet.” Right Hand Man: “I think Gabrielle would be good for that. I’ll ping her on Skype. Andy: “Fantastic. We need to write a dozen job descriptions. Who knows the format?” Suneel: “I do. I’ve written a few thousand of them. I have the Excel template. I always hit the word-count on the nose.” Andy: “Perfect. Now, the client wants to know: six is 12 percent of what number?” Sandra: “Fifty.” Andy: “Very good, Sandra. How did you know that?” Sandra: “You asked it in the last meeting.” Andy: “Good attentiveness and retention. Now, we’re going to have to evaluate and rank about 500 candidates. Do we have the right keyword-recognition software?” Danielle: “I just fed in the latest keywords and it’s ready to go, Andy.” Andy: “Who is good at comparison questions, like “Proficient is to practice as …” Everyone: “Strong is to Exercise.” Andy: “Right, I always forget that one. Can you take those, Esmerelda? Esmerelda: “I’m on it like brown on rice, Andy.” The meeting goes on like this for another half hour, because this is what we need in businesses today — people who can follow instructions . exactly The Bait Crossover for Work advertises all the time, many of them highly paid, many for senior developers and architects. Everyone works from home. They have extensive software systems for tracking employee performance. dozens of jobs Because they pay so well, they get thousands of applicants. Because they get thousands of applicants, they use standardized tests and automated screening exercises to eliminate most of them. I was interested in a job they had advertised — being CEO of an acquired company. This involves using their template and system to run the company, of course. They say this job pays $800,000 per year, so I decided to apply. I have worked for myself for 40 years and have been CEO of a dozen startups. I don’t have an MBA — a universal signal of business competence (except for the fact that only 14 percent of billionaires in the world have one). But I was once a candidate to be the dean of Stanford business school, so I figured I would have a shot. First, I had to take an initial CCAT test, which gives you 15 minutes to answer 50 multiple-choice questions. The questions involve calculations, comparing shape differences, analogies, word definitions, word problems, etc. They tell you to practice using the JobFlare app, which gives you completely different questions and has nothing to do with the CCAT. This is the kind of test it would be easy to do very well on with unlimited time, but in 15 minutes you have 18 seconds per question, making it impossible to get them all right. There’s no penalty for guessing, so you do what you can and guess the rest. There’s no chance to go back after you answer a question. I have studied statistics, behavioral economics, and decision science. I figured I would do well enough to get through this part of the screen. I didn’t. It wasn’t even close. They want you to get 40 or so right. If you don’t pass, you have to wait 6 months to try again. I waited. I still didn’t have work, so I tried again. I realized this is similar to an SAT or GRE test. To win, you must game the system. Crossover CCAT So I went to and paid $50 for their study package. This was an incredibly lame product. It was a very basic page of HTML with some links. Two links gave me a chance to take a real CCAT test without getting my score (that probably costs more). Then, there were five PDFs containing test questions like a real test. It wasn’t a test app, it was just PDFs. You get the question, then the answer. You have to scroll a lot. Some answers have explanations, some don’t. A few of the answers are wrong. Several of the tests have repeat questions. Then there was another test in a different format. Crossoverccat.com I wrote to Mark M (crossoverccat@gmail.com) and told him I was very disappointed that he would charge $50 for such a poor product. He replied: Hi David, A lot of effort was put into producing the .PDFs with the answers, that is the main study material. Everything you received was listed before you purchased it. Why exactly do you deserve a refund after receiving all of the study materials? I said thought it was a very poor product that made the customer do unnecessary work. I asked him for a partial refund. He replied: Out of the hundreds of people the made good use of our product, you are the only one that complained. Thanks for being an ungrateful ass. That’s customer service! So I suffered through the PDFs and memorized a bunch of the questions. Probably about 15 hours of study. I can’t recommend his tests, but you do need practice tests to pass. Taking the Test Again I felt I was ready, so I took the CCAT test again. Fortunately, many of the questions were exactly those I had memorized from the practice test, so even though it was really presented poorly, the test prep helped a lot. I knew the answers and could move on and spend time on the new ones. I must have come very close, because the system said I had almost passed and would be allowed to try again. I tried again the next day and passed. It wasn’t easy, but because I had memorized so many answers, I was able to finish early and still pass. The Essays Since this is a job where you are running a company, they want to see if you have the skills. So they ask you to pitch a business idea in their format. They want an outline, then they want a narrative that anticipates and handles objections. They give instructions and use Elon Musk’s pitch for Tesla as a key example. Then, they want a 5-minute pitch video. Since I know venture capital well, I created a pitch around a next-generation venture product that is very innovative. I spent about 15 hours on these documents and making a five-minute video. I sent them in. Here is their response: Thank you for taking the time to apply to Crossover. We’re sorry to inform you that we can’t approve your application at this time — this means you will not be considered for this position. That’s it. Thirty hours of work, $50 on test-prep, and that’s what I got back. The Switch Do they really pay as much as they advertise? I expect they don’t. I expect no one actually qualifies for the $800k salary for CEO of an acquired company — it just gets a lot of people to apply via their automated system. I would be interested to learn how many of their $200k-and-up jobs actually fill at that salary level. The Reviews On Glassdoor.com, there are over 850 reviews of Crossover for Work. The vast majority of them are negative. They say things like: This company thrives on dysfunction, disorganization, and fear. You are constantly walking on eggshells, constantly having to fight for accurate payment for work you’ve already done, and constantly having to take a stand to receive the most basic things. Pros: None that I know of. Even pay is bad. Cons: Very bad organisational culture. I worked at Crossover for Work full-time for more than 10 years. This company uses people as tools, plain and simple. If you work in the US, and want a real competitive wage, and the ability to think creatively, be rewarded and recognized for your contributions, feel like a real contributor and want other typical things — like health care, 401k, health club discounts … well, this isn’t the place for you. Worksmart —their monitoring tool — if you are a professional, especially an engineer who often has to think creatively to solve problems, you don’t need a “key counter” or screenshots — or even a CLOCK to punch. You need time to focus without worrying if you’ll be paid correctly, not whether or not their algorithm detected you weren’t so productive for 2 hours on Tuesday. Please, use your talents at a company that cares about its products AND its people. They are taking screenshot of your screen every 10 minutes. You can be fired with no reason. … despite being the highest performer on my team I was let go because the bottom line is others on my team were just liked better by my manager. I don’t have any tolerance for those games, that’s why I liked crossover, because the metrics couldn’t lie. But people still can. There is a bleak general atmosphere. I haven’t met anyone who is putting in more work than it’s required of them. Everyone is doing their job just enough to not get fired, which suggests a lack of passion, and it’s due to the way employees are viewed by management. The arrangement is such that anyone can be let go at the drop of a hat, which makes it all the more difficult to reach a state of mental comfort conducive of focus and performance. There is no actual motivating factor other than that weekly paycheck. It’s very dehumanizing to define a person by the sum of their metrics. Basically top management is obsessed with that stupid idea of creating a factory producing/maintaining software products on assembly line and they seem to be ready to sacrifice everything for that ideology. Nobody cares about real result, the only things that matter are metrics and FTAR (first-time acceptance rate). The latter basically means that you’ll be screwed up for any minor mistake like failed code review or some mistake in jira ticket. There is no promotions or any way to show any recognition for a good work you are doing over there, you could be a top performer for couple of months and certain clients ask for your you to handle their issue and still you are in the same place, and to get a higher position you have to go through the tests as if you are an outsider. What I really disliked: Constant tracking/ justification of work stream. Seriously. As others have pointed out, it’s difficult to actually *get* credit for a full work week without working extra. Especially in some of the higher-level, more ‘creative’ positions such as architect, product management, etc. there’s minimal or no opportunity to review or think over things. For me, I work in bursts followed by small distractions in which I’m running the problems in the background of my thoughts. A variety of coworkers and management in my history have almost universally commented about the volume of good work I produce. Even my peers at Crossover had no problem with the quantity or quality of my production. However, their tracking software and systems simply don’t credit anything other than linear, constant “work”. This was bad for me, resulting in me working extra, reworking things as I attempting to change my processes, “faking” it, or simply working longer to attempt to make my hours. I also felt bad for some of the more junior or “factory” positions. It really is tracked by the minute, with lots of incentive to find “problems” with productivity. This is really a thinly-veiled method of wringing blood out of a turnip, by finding flaws or gaps and essentially docking pay. Yeah, the salaries are good but the amount of ancillary work that goes into making “real” hours is awful, and I felt like a chump contributing to it. I had to quit for my sanity. The reviews are worth reading. Some people like the work environment. Some people say they are pushed to be their best every day. But in my reading, it’s really the people who are good at moving things from the in-box to the out-box, not creative or critical thinkers. It didn’t surprise me that the best review was from a military guy. Someone from Crossover responds to many of the Glassdoor reviews, though they are mostly cut-and-paste. Here’s a typical response: We’re sorry to hear the experience didn’t work for you. It isn’t for everyone. The factory model is the best way for us to scale our business when acquiring one company per week. Thank you for the feedback! Here’s before finding a better job: a video from a guy who worked there for four weeks A good summary from an executive: … the company likes to present itself as having a methodology that is fully repeatable and applicable to all of their businesses, but the reality is that this is only a (small) portion. It’s really that they have a bunch of talented people (and that’s for sure) held hostage by the wage/tracking situation who constantly over-perform until burnt out. I believe they could come up with a blended approach, that encourages performance without treating people like cattle (or gears) and maintains a high standard. Given the background of the founder, this is unlikely, as he’s long believed in this kind of philosophy. For me, it was a fun and interesting experiment that I am glad is over. Why do People Work? Management is hard. I always say that executives play golf because it’s so trivially easy compared to managing people. But management comes from a fundamental philosophy, and the fundamental philosophy at Crossover is that people are fungible work units that should be replaced by machines as soon as possible. They are a nuisance to be tolerated as long as they hit their numbers. I wouldn’t fit in at Crossover. I’m not good with rigorous nonsense. I hire employees who ask good questions. I hire people who aren’t afraid to tell me I’m wrong. I want people who are willing to try something new, to fail, to learn. I want people who replace their own jobs with something better. I want people who will break the rules to make customers happy. I want to to accelerate my company. I’d rather coach them and do experiments together than tell them what to do. harness the intrinsic motivation in people Like most people, I need income to survive. But I wouldn’t work for Andy Tryba for any amount of money. I’d rather work for : Frederick Hertzberg or Dave Snowden: or : Paul Akers Or : Jeff Bezos Or : Ray Dalio Or : Bob Sutton Or : Richard Sheridan Or : Leandro Hererro Or : Gary Hamel I don’t think people work for money. I think they mostly work for a) enough money that they don’t have to worry, b) a decent, respectful, and challenging work environment with a chance to make their own decisions, learn, and improve, and c) to contribute to something they find meaningful. That’s why I’m starting the — because too many things in today’s society are broken, and they can’t be fixed by people who are very good at checking the box with the right answer when the other three are already known to be wrong. Because , and Giordano Bruno Institute the world doesn’t work that way culture is critical to success. Trying to control all the variables, including the people, is . the illusion of control My Open Letter to Andy Tryba I’ve paid $50 and 30 hours of my time to say this, because he wasn’t listening. Maybe he will now: Andy, How long do good people last at your company? You can’t measure the opportunities you’re losing because you don’t trust your people and you don’t encourage innovation, creativity, opposing views, experimentation, and emergence. You’re building a monoculture farm, and we know monocultures can be wiped out by a single attack. A purely transactional culture is a graveyard for human flourishing. I am available to help you add joy and purpose to your company, but I’m expensive, and you’ll have to take a few tests . Here’s something I will teach you: Satisfaction is to employees as profits are to growth. Sincerely, David Siegel is a serial entrepreneur in Washington, DC. He is the founder of . He is the author of and . He gives speeches to audiences around the world and online. He helps companies build a more antifragile culture and innovate faster. His full body of work is at . David Siegel the Pillar Project The Token Handbook , Open Stanford , The Culture Deck , Climate Curious , The Nine Act Structure dsiegel.com Also published at https://pullnews.medium.com/applying-for-a-job-at-crossover-for-work-14a79ed37b

My Terrible Experience Applying for a Job at Crossover for Work

Mi terrible experiencia al solicitar un trabajo en Crossover for Work

Online marketing and (EU) elections in the digital age — your choices are not your own On the ethics of Big Data, persuasion, and the existence of free will Can any of us truly maintain that we choose what we choose freely in today’s world? What is unethical manipulation and how can we discern it or protect ourselves from it? On the train to Brussels, as I see the neat, flat and rainy terrain of the Netherlands gradually turn into the slightly more lush november landscape of northern Belgium, a single thought keeps coming back to my mind. Kind of like an annoying wasp. I have been invited to attend the European Parliament seminar on “fighting disinformation in the digital age”, which is right up my alley as a philosopher of marketing and technology. I can’t wait to learn more about the workings and ethics of cutting-edge communication and data technologies. What makes a person choose anything? The question that keeps coming back to my mind is: what really made me choose to come to this seminar? Free will does not exist and big data does Contemporary scientific consensus holds that f exist. The logical idea that determinism rules and that free choice or free will can not exist sits very uncomfortably with me as a human being, a writer, and a marketer. ree will does not So, my genes combined with my life experiences made me choose to write this article? But what if I didn’t want them to? If free will indeed does not exist, then how bad can it really be that marketers and political campaigners alike use Big Data and methods like microtargeting, psychological profiling? Or that they deploy bots posing as (human) social media accounts, neuromarketing and other advanced science and technology to sway people in their preferred direction? One of the main reasons I was interested to see what the seminar in Brussels was all about, was that Jeroen van den Hoven, professor of Big Data Ethics at the Technical University of Delft (TUD), would be present during a panel discussion on the second day. So I would be able to ask him precisely that question. At least, that’s what I thought my reason for choosing to attend was. But who knows? Why digital marketing techniques matter to the EP A large part of the seminar was dedicated to explaining why the European Parliament and European Commission have been closely watching elections around the world — including the ones surrounding Brexit and the infamous Trump election, but also in the eastern European region, South and Central America and Sweden. And, how the EU is learning and analyzing what effects disinformation and innovative communication technologies can have on the outcome of elections. The trend is of course alarming, especially considering the crisis of politics EP vice president Pavel Telicka reminds us we are currently seeing the consequences of around Europe. Throughout the continent we’re seeing a rise in populism and a wave of local nationalism, which are both exacerbated by the rise of fake news or ‘misinformation’ as the EP members prefer calling it. Even in the 2018 local elections in the small state of The Netherlands, where I live, we learn that , mostly to the benefit of Geert Wilders’ right-wing PVV. both (possibly illegal) microtargeting and twitter Bots were used What the EU is doing to counter misinformation The EU has a number of projects and task forces dedicated to solving this complex problem. Among them are (panel for Science and Technology Options Assessment) and the initiative the EU helped set up that resulted in a self-regulatory , signed by Facebook, Mozilla, Google and Twitter among others and directed at fighting fake news online. STOA code of practice There’s even talk of raising taxes on online advertising to a level that would be comparable to the level of taxes on offline advertising. This would indeed be a very bold move, and I am curious to see how seriously the EU is thinking of making this a reality. But with the coming into focus, the aim of the seminar on fighting disinformation was clear. What the European federal government wanted was to reach out to journalists and voters alike, to crank up the debate about misinformation and its effects on the political climate and elections. EP elections of may 2019 EU to use Blockchain, AI in the fight against disinformation? When we speak of discerning the correctness of information provided to us, we’re essentially speaking about trust and the robustness of records against tampering. If it’s anything the much-hyped blockchain and other forms of decentralized technology could bring to the table, it would be . solving for trust One of the core ideas of Nakamoto’s brain child was the immutability of records that are always public. Additionally, v arious blockchain initiatives have been associated with the countering of fake news in the cryptosphere. Is the European Parliament looking into using decentralized technology in the fight against fake news? Eva Kaili, MEP and STOA chair, tells us that in her opinion Blockchain and decentralized technologies are as yet not mature enough to use for problems of this complexity and scale. However, she admits the philosophy behind them is certainly very interesting and possibly even revolutionary. The EU is currently looking into use cases for blockchain, potentially combined with A.I., for fighting disinformation, equalizing the media and advertising landscape and for safer use of data and identity. What we ourselves can do about misinformation and technologically aided manipulation Carl Miller, and author of , tells us a marvellous story about being a ‘Fake New sMerchant’. He illustrates what an amazing economic opportunity it can be in regions of Eastern Europe where youth unemployment can be as high as 60%. His plea is that we ‘not only burn the poppy fields, but also give them something else to grow’. research director of CASM at Demos Death of the Gods After Millers inspiring talk, we learn more about what we ourselves can do to fight fake news, both as journalists and informed citizens. Eoghan Sweeney from takes us on a tour of their free online educational resources, but also a lot of free online tools and their usability for “Intel Techniques” that we can all apply. Among them are staple apps such as Google Maps, Google Earth, Russian Yandex and Wikimapia, and a fantastic app by the name of ‘ ’, which lets you determine what direction the shades in a given photo should point to if it was indeed taken at the stated time and location. Firstdraftnews.org Suncalc The main thing, though, is not tools or the use of them. It’s ourmindset. The main point is that we firstly raise awareness about mis- and disinformation amongst ourselves, and secondly adapt a thoroughly critical way of thinking about information we receive in any way, shape or form. And, maybe, that we realize the existence of our online and offline filter bubbles. Our digital ecosystem and responsibilities for business What role does the notion of ‘free will’ have to play in all this? On the second day of the seminar Professor Van den Hoven wonders if ‘Democracy will survice Big Data and A.I.’. He argues that we must all take this issue much more seriously than we have thus far. People and organizations that use technology like Big Data and A.I., combined with others such as (political) microtargeting and combined with expert knowledge on psychology and psychometrics, can achieve much — even to “devour human autonomy”. Quite the bold statement. It can be argued that marketers and business also have a responsibility for the misinformation mayhem we’re experiencing as a global society. I ask the professor wether he sees a responsibilty or a role for businesses in fighting misinformation and manipulation, considering businesses are the paying customers who helped create and sustain our current digital ecosystem. The main concern Van den Hoven raises is the following: there has been and continues to be a strongly asymmetrical build-up of knowledge about persuasion and the use of technology in it. Both politicians and businesses know vastly more about them than the general public, allowing them a huge and unfair advantage. This needs to be regulated, the professor states, and I agree. But, why wouldn’t a sort of self-regulatory initiative be set up where parties and organizations take responsibility into their own hands? Transparent communication and the ethics of influencing ‘Free will’ is a complex and vague construct at best. The same could be said about persuasion, an obviously related concept. But regardless your personal stance on the matter — believer in free choice or not, you decide — at some arbitrary point we must agree that the providing of information to facilitate a person’s choosing ends, and unethical manipulation begins. Where does Van den Hoven stand on the matter? According to the professor of Ethics & Big Data, there is a fairly decisive line of division between convincing and manipulation or brainwashing. He illustrates his point with the legal definition and treatment of what is called ‘entrapment’. If someone is put in a certain position, with a certain level of cues and coaxing to commit an illegal act, we as a society rule that they can not be found guilty of choosing to commit the crime. In the same sense perhaps, we can not think that a voter or consumer can be viewed as ‘freely choosing’ what he chooses, when we use some of the techniques and technology being used all over the world, today. Transparency, and the active propagation of ‘Information-Awareness’ is what is owed to citizens, patients and consumers. Beyond GDPR On my train ride back to The Netherlands, my mind is still — or even more — dazed by the . riddle of free will But my resolve is strengthened: I intend to double down and to learn more and talk more about the ethics of the use of Big Data, . and the immense impact digital communication technology has on all of our lives I already talk about these subjects to my friends, readers, and the people that I work with in marketing and tech. Sometimes to their wonder and inspiration, other times to their boredom and indifference. It is somewhat tedious subject matter at times, surely, but I think maybe that should be a price we want to pay for our freedom. When the GDPR ‘hit’ our marketing department, I was the one advocating we go beyond the law, and stop to seriously think about the impact and responsibility that we have on people’s lives with our own digital marketing practices. Even if that meant generating less leads. We, too, used to make the use of Facebook and Google pixels a standard in all of our websites, I’ll admit. Maybe we should all go beyond the GDPR and the constraints as well as the awareness it bestows on us. As politicians, businesses, and individuals alike. To protect our own and each other’s ‘free will’. You decide. <a href="https://medium.com/media/3c851dac986ab6dbb2d1aaa91205a8eb/href">https://medium.com/media/3c851dac986ab6dbb2d1aaa91205a8eb/href</a>

Online marketing and (EU) elections in the digital age — your choices are not your own

Everybody loves the concept of mentoring. You will both help your padawan and clarify things for yourself. Maybe you will also get a promotion for such activity. So, there are no drawbacks at all and if you have a chance, you should try it. But hard-hearted senior developers usually forget: a junior is shivery and vulnerable. He is almost like a child who needs to be taught everything in this unfamiliar world. And his further career depends a lot on his mentor. I said goodbye to my last mentor a few months ago, and I had been almost under continuous care before that. I had five teachers. While my experience is fresh, I want to share it and warn those who are passionately willing to learn from irreparable mistakes. Or just publish a bullet list for juniors to check that your mentor is not an asshole. Teach to fish is better than give a fish Do not give final results “correct” answers. Point out the problem areas and give tools to solve it. Each final list of decisions that you give to a person will become obsolete over time. You need to create an educable developer, not an educated one, who is able to find solutions, not to memorize the standard set. It is more important to teach tools and principles, rather than specific solutions in the code. , show how you use IDE, debugger and a version control system. There are many obvious optimizations of your workflow, which a junior will get only after a year or so. Save him some time at the start. Try pair programming I did not immediately figure out the . In the beginning, it seemed to be a good idea for me to just press and pray. At some moment, this did not work, and I scribbled to my mentor. He called me in Skype, I shared the screen and like a monkey pressed the buttons, which he quickly dictated. It worked. The next time the situation was the same. When the problem has repeated for the fourth time, I finally guessed to read the guides. Interface Builder “Reset to suggested constraints” “not working, needs help” Shitty fish-rod is worse than nothing If your ward respects you, he will not doubt your words. Therefore, take responsibility for your pieces of advice. Affirm something, just being absolutely sure that there are no alternatives. Explain why this is the best solution, which advantages it has, what kind of problem it solves. It is better not to give a solution at all, but to list methods known by you, their pros and cons. Let junior learn to take responsibility for decisions himself. During analyzing and comparing alternatives he might teach you something occasionally. When it came to network requests, I obviously used . Sometime later, my mentor looked at the code and was not very happy. His final advice was to replace the code with synchronous . After call, the thread was locked and unlocked only in the task callback. No comments. Alamofire NSURLSessionTask .resume() He also shared his own implementation of type and told me to use it without further explanation. The idea of , as I understand now, is ok… But that time I just left this file to lie and die in the project directory. Even one attached article about the could greatly boost my understanding. Result Result Railway pattern Tired, cynical and bearded You are not a coach on personal effectiveness, although you will inevitably have a strong influence. At least you should do no harm, nevertheless, even minimal motivation is going to be useful. You are like apostle Peter at the Pearly gates. Try to present IT in a slightly more favorable light. It is clear that you are a professional who fixes bugs with crutches 24/7, but … Share IT celebrities twitter, or have a beer on local devs meetup. Or just recommend , it might be quite exciting. Show a fun part of our sphere, not just dull days of dragging tasks in Jira. jokes from large companies Swift by Sundell Be less serious in face-to-face meetings, do not show your fatigue and pessimism on a newcomer who has not seen or achieved anything yet. Encourage any development, even if you have already fed up. Junior wants to try a new framework? No problem, go. Interested in linter setup? Yes, we have always dreamed about that! Interested in Rx? Here are a few useful articles about it. It is clear that strict business wants it “to be done tomorrow”. But let him spread his wings before a real production. Junior is a long-term investment, you should not try to get a profit from him immediately. My mentor has always listened to me and often approved technical solutions I had found. I will always respect him for letting us throw out legacy code when someone else would cry: Sometimes he learned a new solution from us and we had a long discussion (just like seniors) about its convenience. It was obvious that he cared about our professional growth. “Oh, this has low priority now, don’t touch it.” Here is not there Attention is good, but your time is also required. You can’t be authority just because you are the teacher. You should dig into junior’s concrete tools and current tasks. It is always more interesting to learn from a person who gives a shit about you at least minimally. You should not break the initiative of using technologies unfamiliar to you. If he can justify his choice, let him go. It does not matter that you do not have experience in the selected framework. Your experience allows to understand it much faster. , and — these phrases should be excluded. You are the king and god only while you quickly and correctly answer questions, solve problems that appeared. If you are only able to come, moan five minutes, “turn on / off“ and leave, then you are not good at all. “I knew, but I forgot” “I haven’t used it for a long time” “I have another tech stack now” My app crashed due to an unconnected . This can be recognized in a second if you have ever encountered a problem. My sensei was sitting, swearing, cleaning for an hour. As a result, the decision was suggested by some intern. outlet derived data Clear growth path Junior will inevitably suffer and fuck up. After all, this is the “experience”. Make his complicated path easier. Provide task choice and clear acceptance criteria. Build processes like in real production world. Let it be sprints and planning, grooming and retrospectives. This will create an opportunity to regulate a schedule. Junior will be able to choose tasks that he is interested in or which he understands, how to do. In case of stupor or long waiting for an important answer from somebody, he can just switch to another task and not to languish. And of course, there is motivation of the fulfilled plan. Did you complete the sprint plan in the middle of it? Well done, take a beer and . sing ACDC song My backend team has sat without tasks for two weeks because: . Frustration and cannibalism in the ranks of developers. Always feed your beasts. “frontend is lagging behind, we must wait, do something for yourself” Are you surely a senior? If you are just a dude who sometimes comes out from nowhere in a crumpled shirt, the impression of you is disgusting. Your old colleagues know that you are a cool reviewer, excellent architect and once shared a bottle of an expensive whiskey in the office. But junior can’t recognize good/bad programmer. Everyone is equally cool for him. study hard and become me Think about your representation. Show that you are doing something. Make him want to get into your life. Order him to subscribe on your Instagram, where you post photos from conferences and meetups, or show a podcast episode with yourself. Share a link to the blog, show a certificate. If there is absolutely nothing to boast about, maybe it’s too early to teach someone, isn’t it? “employee of the month” Talk more to junior. Become an idol for him, let him think that you are doing something beyond the limits, even if it is not true. Make him respect yourself. And be near, let him see what you do, that you are not somewhere in the clouds, but with him in the same company, working on a similar project. Be alive and involved. Show that you have interests, that your growth is not over. Because soon junior will become an old fart with his family and mother-in-law, and he will definitely have no time for development and knowledge. And now it is his brief moment of productive life. My mentor was just such a dude, who comes to the office once a week, flops on the sofa, sighs and says: With all my naivety of that time, it was impossible to believe in his competence. I did not want to evolve in order to eventually become similar to him. “Well, what have you got there again?” I will grow up and become like you clap if u remember those times My friend junior told me a very sensible thought: “Now I finally see the path to become a person I want to be. In order to go through this path you only need diligence and will. And talent doesn’t matter at all” A beginner sees a rockstar programmer who turns in the middle of a development process, gives instructions, offers solutions. Junior looks around at his small tasks and does not understand how to become so cool. Should he just wait for required three years of experience? Tell him about the past, how you started. Stories about successful well-known businessmen are much less impressive than the story about a guy from your university who get an offer from Facebook. Give sincere hints concerning your path to success, otherwise junior will come to frustration. It causes thoughts that he was just born “without talent”, it is not possible to become “like you”. Beware of idioms like I did this, then I read that, then I got there. A two-hour conversation with beer (junior should pay) can give him an excellent development roadmap for the next few years. “well, I always took more responsibility and was not afraid of new tasks.” For just one imperial stout brought, you can hear from the mentor a story about his youth. Hear his/her fate, the motivation of the steps taken in life, and then evaluate the results. And this is not well-known Rather, . “learn from others mistakes”. “cut the path where the tutor went round and round” What are we going to do tonight? There is a lack of tasks for junior level sometimes. The standard solution is to wave aside with words: “ The truth is that “getting familiar” with the code base without a specific task is a waste of time. How do you even imagine it? Just open and read the names of all 1000 project classes? get familiar with the code base.” Come up with something. Do not let the energy boil out just like that, otherwise junior will become prematurely a cynic who is disillusioned with everything. Surely there are unwritten tests or framework whose version you need to update. As a last resort, convince him that it is extremely IMPORTANT to learn Haskell or take a course about ML. I remember at the beginning I had a task to write tests with the , and then a simple frontend for monitoring with . I will never forget how we configure a connector for full-text search worked with . Most likely it was not necessary for anyone at all, but how damn fun it was. Mockito React Elastic Search MongoDB Strict coach padawan, why u pushed to master again Of course, there’s no need to protect junior from all dangers of IT world. He has to be prepared for the inevitable stress and despair. We all know Dr. Cox and . Control junior’s self-respect. Sometimes it’s ok to remember him that he is still not so good. The main thing is not to overdo it, because there are problems with toxicity in the IT community. The ideal option is to make such a team so that every person has someone to level with, try to catch up. “Miracle on Ice” Try to shake it all the time. Come to work and say: Prohibit him to stop even for a second, it is too early for rest. “I started learning a new framework/course/language, it’s so interesting, and you are still sitting, watching TV series?” But do not forget the contrasts. Shout, criticize, and then on review and cheap beer on Friday at the office. Give him an understanding that satisfaction can be obtained not only from money but also from the process itself. “Cool decision, guy” Become an IT bro You have got used to slang and you know the whole community. Junior looks at all this with wide-open empty eyes. Let’s be frank, he can read the documentation of the language and search through StackOverflow. Give him the ground under his feet, give concepts that he can not google or read. Actual answers on questions like , , which are really hard to get. “Who is now on demand at the market?” “How to grow up in a company?” “How to endure obscurantism and shit in the work team?” — Recommend podcasts and blogs. Junior is a vase into which you can infinitely pour, so the extra will go away by itself. But leave the vase empty is not an option. Let him live by programming, breathing with terms, echoing sentences. The sooner he enters the world of Medium, Twitter and blogs — the more competitive he becomes comparing to his comrades. Nobody has explained to me what to say during a stand-up. and I seriously went into a detailed story about which I was testing today. I gained the most valuable knowledge “ only after a year. “What did you do today?” class Always put the smile at the end of the chat message, otherwise, you will be called toxic” I learn a lot from my tutors, they were the ones who give shape to me as a developer. Even bad examples gave me knowledge of how NOT to do, although I would prefer to have only good teachers. After this article, even a junior will be able to adjust the mentor’s behavior and optimize his education. Separate greetings to Ekaterina Petrova and Rustam Kadyrov. They gave me faith in a bright future and pushed me to many important actions in my life. And I will always remember our discussions and evenings in an office. The relationship of a student and a teacher — the most exciting thing for the whole career. It is always nice to have a moment of nostalgia for cozy minutes of collective working ecstasy. After all, close friendship is mostly born from such a relationship. And friendship is magic! Respect your juniors, grow them free from toxicity. Take a short survey — Your opinion really matters to me! https://goo.gl/forms/StARJasrqOpIqiwv1 If you liked the article, give it a clap (50 times) and share it. Subscribe me to read more about iOS development. Provide your children with normal mentors :)

SUNCEL

suncel.io

-- employees

Since n.d.

COMPANY RANKING

SUNCEL

EVERGREEN INDEX #10695

Representing this company? Click here to claim and customize this page!

Suncel WIKI

Apply to get your Tech Company News Page live on HackerNoon.com today!

SUNCEL

suncel.io

-- employees

Since n.d.

COMPANY RANKING

SUNCEL

EVERGREEN INDEX #10695

RELATED COMPANIESRANK

STORIES ABOUT SUNCEL

Representing this company? Click here to claim and customize this page!

Suncel WIKI

Apply to get your Tech Company News Page live on HackerNoon.com today!