Geoffrey Negiar

"There is a lot of joy to be found in one's craft and a good engineer must be a craftsperson, dedicated to learning new things, understanding problems and tools deeply."

Geoffrey Negiar is a machine learning engineer based in Northern California. He is currently a PhD student at UC Berkeley and holds a master’s degree from ENS Paris-Saclay’s Computer Vision and AI program (MVA) and is an alumni of the Ingénieur polytechnicien program at Ecole Polytechnique. Negiar’s research focuses on designing efficient optimization algorithms, modeling in the presence of uncertainty, and natural language processing. He has interned at Bloomberg LP, Shift Technology, SumUp Analytics, and is currently interning at Google Research. He is also a member of the French Armament Corps (the French DARPA Fellowship).

In this interview, Geoffrey Negiar shares the differences between studying in the United States and France, when he realized he could use math to make a social and environmental impact, and his perception of STEAM versus STEM.

TGL: How was your childhood?

GN: My childhood was happy and formative. My parents were both relaxed and very energetic. I grew up in their world, not having any siblings. In the late 90s, when I was 4 or 5, my mother launched her brand of Japanese green tea in Paris, one of the first in Europe back then. She was originally from Los Angeles, so we bounced from Paris, where we lived to Los Angeles to Shizuoka. I feel at home in all three countries, and I miss the other two dearly when I spend too long in one of them. I was definitely an American kid growing up in Paris, but through my Japan trips and martial arts practice, I feel like my current way of life is heavily influenced by Japanese Zen, combined with a tension between the American "make it work" spirit and French abstraction. My father picked up a lot of random hobbies, which has had a large influence on me. I also love to pick up new crafts, and try to understand them from the perspective of other paths I have walked for a while. There is a lot of joy to be found in one's craft and a good engineer must be a craftsperson, dedicated to learning new things, understanding problems and tools deeply. Cooking, bartending have helped me more than I can describe for the machine learning research process.

TGL: You recently recommended the book The Innovators by Walter Isaacson. Why do you recommend this book?

GN: The Innovators gives a long term perspective, starting with Ada Lovelace, the daughter of Lord Byron, on the rise of the digital age. Isaacson shows how different personalities set in specific environments interacted, maturing the technology and the mindset that allowed computing and networking to permeate today's society. One point struck me in Isaacson's analysis: "The truest creativity in the digital age came from those who were able to connect the arts and sciences. They believed that beauty mattered."

TGL: You studied at Polytechnique in Paris and UC Berkeley in California. What did you study, and how did you perceive the differences between your field in France and in the United States?

GN: I started my studies in France, in prépa, then at Ecole Polytechnique. I first studied broad theoretical basics in Mathematics and Physics, but also Philosophy and Literature. Little by little, I realized the impact that Machine Learning could have, by making obvious patterns in large amounts of data that were not collected until recently. Healthcare, transportation, politics - the Facebook newsfeed! - are obvious applications. I decided to concentrate on this field. This led me to the MVA Masters at ENS de Cachan, now known as the Paris-Saclay, where I discovered the field of Optimization. Given my love of the beauty of mathematical theory and modeling, and that Optimization is one of the bigger bricks in technology since the 50s, this was a perfect combination. I was admitted to UC Berkeley's PhD program in the BAIR, Berkeley Artificial Intelligence Research, lab, under the advisorship of Laurent El Ghaoui.

TGL: What are you currently working on?

GN: I’m working on my PhD degree at UC Berkeley. I work on Robust Optimization for Machine Learning models. Data in general is uncertain: people make mistakes during collection, sensors are not calibrated exactly. Therefore, the numbers you get in your database are not exactly representative of the real life quantity you want to study. We're building a theory to take this into account for different widely studied problems.

TGL: When did you know you were on the right path?

GN: My last year at Ecole polytechnique and the MVA were game changers for my decision. I was always interested in how technology has an impact on people, on local or national cultures. When I started my graduate studies at Polytechnique, I got interested in urban planning: public transportation and walkable cities have a huge cultural impact - it's obvious when you compare cities like San Francisco, Los Angeles, New York and Paris. I realized that "platform planning" is just as important in people's everyday lives: the impact of products like Facebook, Google, Uber, and Amazon is huge. I realized I could combine my love of theoretical maths and modeling, and still have a social impact by delving deeper in this circle, which led me to where I am now.

TGL: STEAM fields are science, technology, engineering, art and mathematics, and for a long time, we talked about STEM without the A for art. How do you perceive multidisciplinary in your work?

GN: Being multidisciplinary has always been important to me. My parents never worked in a single industry for more than a few years: they were art dealers, had a bagel brand in Paris, and are leaders in the Japanese tea business. The industry or field itself doesn't matter as much as committing to delve deep - combining ideas from different fields is tremendously helpful. I loved the program at Polytechnique for this: on top of our engineering education, we had mandatory humanities classes, including a course, "Can one still love democracy today?" by Michael Foessel. STEM is unnatural to me: why would one only focus on engineering and forget the arts and humanities?

TGL: You have a passion for art and events. How does it play a role in your work?

GN: Although the process of research is a creative one, it also has its more tedious periods. The process is similar to the one of art creation: in the end, we're all getting things done and adding our stone to the overall construction, depending on what excites us personally and what the trends are. Staying with people of your field, you sometimes lose track of that excitement. Meeting other creators and seeing tangible, material work is very refreshing and often yields surprising new directions. Events impress me - unlike many engineering projects, where you can iterate on until you converge to something that works, you only have one chance to get an event just right.

TGL: What are Silicon Valley’s weaknesses?

GN: In my mind, Silicon Valley's weaknesses are mostly geographic: the Valley is quite isolated from the rest of the US, and in particular, from places where tech is not dominant - such as Los Angeles and the political power in Washington DC. Getting around in the Bay Area is also difficult; it's great for focusing on work and getting things done, but I also need to get out of the area every so often to gain perspective and refresh my ideas.

TGL: Who would you like to have dinner with that you don’t already know?

GN: Miyamoto Musashi. He was the ultimate walker of the Path, a master of his craft to the point that he let it go. John Von Neumann comes close second, from his contributions in all areas of science both theoretical and very much applied. More realistically, I love to dine with people who can teach me new things, masters of fields I don't know; I love to trade insights on our fields, and understand how they are the same, and how they are different.

TGL: What advice would you like to give to The Genius List’s readers?

GN: Stay curious. Spend at least a few minutes every day on things you want to get better at. Be patient: efforts compound. Find people who are good at something you want to learn to teach you. Ask questions, and watch.