TRI Be Yourself Spotlight: Meet Igor Vasiljevic

At TRI, one of our values is “Be Yourself,” which encourages team members to bring their full selves to work every day. Our diverse views, history, personality, character, and experiences are invaluable to the work we do at TRI and Toyota.

National Intern Day is right around the corner! As we recognize and celebrate the future leaders of the world, TRI is proud to celebrate the amazing work of its interns. Meet Igor Vasiljevic , one former intern who has now gone on to work at TRI full-time.

Igor Vasiljevic

What was your role as an intern? What is your role now? 

In early 2020, during the fourth year of my Ph.D., I had my first opportunity to work with the TRI Machine Learning (ML) team as a computer vision research intern. At the time, TRI ML was doing some really exciting work at the intersection of self-supervised learning and 3D vision (depth estimation). In collaboration with the team, particularly my mentors Vitor Guizilini and Rares Ambrus, we developed a novel method for self-supervised depth estimation that could adapt to any camera type. It was a fun challenge to find videos captured in "unusual" settings to test our method - eventually, we used videos from underwater drones and GoPro fisheye cameras recording off-road biking, demonstrating the broad applicability of our approach.

The following year, I had the chance to come back to TRI for a second internship to work on extending these projects. The work I did in collaboration with my colleagues became the core of my Ph.D. thesis. Upon graduation, I joined the ML team as a research scientist, and I am currently involved in our ongoing efforts to develop embodied foundation models for robotics and beyond.

What inspired you to get into your field?

As a teenager, I tried (and mostly failed) to create my own games, but I learned quite a bit about computer graphics and programming in the process.

My interest in machine learning originated from some of the classics of science fiction - works by authors such as Asimov, Stephenson, and Gibson that explored the potential of artificial intelligence. Among these, Greg Egan left a particularly strong impression on me. His books like "Diaspora" and "Permutation City," exploring the implications of intelligent software, were remarkably ahead of their time despite being written in the 90s.

It’s National Intern Day soon, what advice would you give to current and future interns? 

I’d suggest that current interns try to meet as many people as possible, both within but also outside of your divisions (some of the most interesting discussions I had as an intern were with interns outside of the ML team working on projects that were far outside of my area of expertise).

For prospective interns, my advice would be to showcase your work as much as possible. This might mean presenting your work at conferences, sharing code online, or publishing well-documented project pages.

What is your most memorable moment or biggest takeaway as a former TRI intern? 

My favorite memory as a TRI intern was joining the Tahoe offsite retreat in the middle of my first internship.  As someone who really enjoys the outdoors, it was great to see my team share these interests and to get to experience the close-knit culture of the TRI-ML team firsthand.

What do you like to do for fun? 

I spend most of my free time searching for surf spots up and down the coast, usually somewhere between Half Moon Bay and Santa Cruz.  In the past, I’ve even had the opportunity to mix conference travel, presenting our research one day and surfing the next. I did this most recently at the European Conference on Computer Vision in Tel Aviv – I was very surprised to see surfable waves in the Mediterranean!