In Conversation with a Former NASA Machine Learning Engineer

For most of Kuba’s life, he believed that his future was in mathematics. The grand scheme was to pursue a career in mathematics by earning a bachelor’s, a master’s, doing a few internships, then, eventually going for the big Ph.D.

Well, plans sort of changed.

Kuba explained that this happened during his final year of high school in Warsaw, Poland. Something clicked for Kuba when he took a few MIT OpenCourseWare programs, a web-based publication with free online course materials directly from MIT.

“I was exploring what undergrad mathematics looked like because I was kind of done learning all the high school [content],” he says. Kuba had gone through around four to five courses when it hit him.

“In mathematics, you have to study incredibly hard for around ten years, and then be very lucky with your choice of investigation,” he says. “Then, if you’re very good and very lucky then maybe you’ll discover something that’s new, valuable, and interesting.”

That’s when Kuba decided that even with his extensive knowledge of mathematics throughout his high school and extended studies, he wanted to set a new path for himself, one in computer science.

With the exponential growth of accessibility in computer science over the past decade, it was easy to find coding classes or software development boot camps. But for Kuba, it started from his interest in video games.

“I tried to make my own video games, and that seemed pretty fun, it seemed kind of like the natural thing to study instead,” he says. “It was very self-taught, mainly online tutorials [of coding] and a lot of YouTube tutorials.”

During his final year, Kuba would apply to the University of Cambridge where he would complete his bachelor’s degree in computer science. Soon, Kuba would venture into his master’s degree in 2019 at the University of Oxford, completing his thesis on 3D rotation invariance in deep networks for point cloud processing.

Kuba has an incredibly extensive background in engineering, spending time at Google as a software engineer intern in 2019 in Switzerland. In 2020, Kuba would go on to intern at NASA as a machine learning engineer.

“When you grow up and watch American movies, you hear about NASA a lot, and it ends up being this idealized place for science,” he says. “At least that’s how it seemed to me.”

While Kuba may not view NASA as his dream job, there was something intriguing about the organization that kept him applying. “During my undergrad, I was rejected, but then I applied again during my masters…and it worked out.”

During his time at NASA, Kuba would be matched with a professor with a focus on fluid dynamics. Here, Kuba was able to explore and learn a lot about physics and was able to apply his machine learning skills from his university experience.

“Interdisciplinary research is just a new different class of exciting in my opinion,” he says. “It’s collaborations where no scientist can solve a problem on their own, but a group can make it work together.”

By the time it came for Kuba to apply for his Ph.D., his decision changed after meeting one of the co-founders of Cohere, a company where Kuba is currently a machine learning engineer.

As Kuba describes it, one night, during his master’s at Oxford, he was introduced to Aidan Gomez at a social gathering, who at the time was completing his Ph.D.

“At some point, I stumbled upon one of the papers he is most famous for, which is one of the cornerstones of modern AI and natural language processing,” he says.

“So I reached out to him, and was like ‘hey, want to talk?’ I was under the impression that he was happy to give some career advice…I met up with him in the college cafe, and the career advice ended up with ‘I have this company.’”

That company Aidan spoke of was Cohere, an AI startup whose natural language processing software provides an improved quality of understanding human language, building accessible conversation between humans and machines.

From intern to machine learning engineer, Kuba’s journey to where he is today has helped him grow and gain knowledge that aids him as a machine learning engineer. When asked whether he has advice for future generation computer science students, he emphasized the value of learning.

“If you’re a theory inclined student, you might spend your first year of undergrad learning and never really having this other mindset of trying to make something, which is a necessity for companies, that’s how they make money,” he says.

“It’s definitely a good idea to strive for variety because that way, you’re giving yourself the best chance to discover your ideal kind of work environment.”

For Kuba, this meant interning as a video game software engineer at Kythera AI, an academic research assistant at EPFL (École Polytechnique Fédérale de Lausanne), and a software engineer at Google. Balancing both academic and industry internships throughout his undergrad has given Kuba valuable insights and a clear direction for his future plans.

“In smaller companies, there’s the argument that you’ll have more impact, and maybe you’ll be more flexible on what you’re working on, especially with really young startups,” he says. “But it’s always good to have a [big name] on your CV.”

Kuba’s plans from mathematics have transformed and thoughtfully scaled into a specialization in machine learning. It may not have been exactly what he planned for in the beginning, but plans change with research and real-life experiences.

He recommends to students with similar ideas on career projections to always be prepared and not be afraid to step back from the original goal to learn new fundamentals as you go. And who knows, you may be right back on that same track later on in life but will be better equipped and wiser than ever before.

Also published here