ML@GT Alumni Corner: Jason Lin, Seeking Balance between Research, Application, and Problem Solving on the Go

The Machine Learning Center at Georgia Tech is responsible for training the next generation of machine learning and artificial intelligence pioneers. Jason Lin, a recent master’s in computer science graduate who specialized in machine learning and robotics, is one of many students the center is thrilled to see thriving in today’s workforce.

Currently a research engineer in perception at Lyft Level 5, the company’s self-driving unit, Lin also has had experience at companies like Google, Spotify, and Apple. Passionate about hackathons, Lin has developed tech for social good and placed in numerous competitions like Stanford TreeHacks and Facebook Hackathon Global Finals. Lin also enjoys attending conferences and contributing to open-source projects, and we’re delighted to discuss with him about his experiences, both the good and the bad, in the hopes that it will help a fellow Georgia Tech graduate find success.

Hometown: Calgary, Canada

Advisor: Frank Dellaert

Unknown-1 — Lin previously interned on Spotify’s security team at their global headquarters in New York City.

Tell me about your research interests. Where might people be impacted by them in everyday life?

My research interests are in deep learning interpretability, fundamental visual, and 3D understanding, with an emphasis on perception for autonomy and human-centered AI. The implications of explainable computer vision are ubiquitous, ranging from robotics and augmented reality, to healthcare, retail, and every aspect of how we interact with each other and intelligent machines.

You’ve interned for a lot of major companies. What has been the best lesson you have learned from your internships?

To put it simply, be humble and stay hungry. Always be proactive to learn new things and acknowledge that you may have to adapt to different environments, work on various projects, and communicate with different teams to be successful. It’s important to stay laser-focused on one’s projects and fulfilling the team’s needs, where dissecting ambitious objectives into measurable milestones helps to alleviate much complexity. Set out an expectation or internal compass of what you‘d like to achieve at the end, but also don’t overcommit. If it helps, I also like to ask myself if there’s any additional value I can bring to the table and seek to go above and beyond from there.

Unknown-3 — In 2017, Lin and his teammates won third place at Facebook’s Global Hackathon finale.

Any advice for students on finding the right internship?

Prioritize project alignment, mentorship and the capacity to grow, such as the skills you’re projected to learn, and where one can have the most impact. Location may be a priority if you’re evaluating where you want to work but don’t let compensation or company prestige keep you from an opportunity that’s right for you. A good indication is going to a place that’s ascending and have interesting work. Finding the right internship is like a mutual match, and it can be through different venues, so put your name out there. I had the incredible opportunity to network with an alum at CVPR and be recruited to Lyft.

What advice or words of wisdom would you give to fellow students or colleagues who have to move around frequently for various opportunities?

Instead of viewing the frequent travels as an inconvenience, welcome it as an opportunity to learn about fresh perspectives. When I was living in New York, I learned that living in a city feels totally different than visiting it. Observe, adapt, and be ready to thrive in a new environment by always staying open to change. In terms of day-to-day organization, I like to refer to the KonMari method, where I keep things that I either need or that spark joy in life.

Lin aims to work for companies whose core values align with his own. He embodied one of Spotify’s core values – playful – at a private concert.

What is it about machine learning that excites you?

I’m excited about its interdisciplinary nature and real-world applications in traditionally isolated domains like agriculture, athletics, medicine, etc. I also enjoy the rapid pace at which the community advances, meaning there is always something to learn and improve upon. As ML becomes increasingly accessible through resources made public, it will help free people from the burden of repetitive work, leaving us room for creative tasks that catalyze future breakthroughs in science and the arts.

Lin enjoyed his time as an AI Resident where he worked on applied Machine Learning at Google X, the Moonshot Factory.

Why do you think embodying Georgia Tech’s motto of “Progress and Service” is important, especially in regards to artificial intelligence (AI) and machine learning (ML)?

The emergence of AI/ML catapults us into a new era of productivity, but with greater power comes greater responsibility. Recent work at Georgia Tech has shown that bias in machine learning could exclude under-represented members of our community, thus it is important as practitioners to be reminded of our purpose to improve the human condition and to be inclusive of all.

While AI’s automation potential is widely acclaimed, its potential for societal benefits i.e. access to personalized education is often overlooked because of the lack of lucrative profits and diversity of creators. We’ll need to think about AI from the bottom-up: start with a problem, and then look for the solution. We shouldn’t only be developing technology for novelty’s sake, but also think about how we can leverage AI’s scale to address a massive gap in economic parity and individual creativity.

What advice would you give to current Georgia Tech Ph.D. students?

Progress and success come with consistency. While the journey may be rough and peers may dwindle as you continue to push human knowledge forward, know that every day you’re working towards being the very best expert of a field.

That sense of fulfillment was what motivated me to invest extra effort in early research opportunities, which over time accumulated to make a difference and prepared me well for future endeavors. Stay curious but also specialize, and go deep. While you’re at it, try to foster connections between disciplines and develop a work ethic conducive to experimentation.

Unknown — Lin presented his work with co-authors Alex Cabrera and Fred Hohman on interpretable machine learning during a demo at CVPR 2018.

How has your time at Georgia Tech impacted you after graduation?

My time at Tech trained me to be organized and systematic with the way I approach things, i.e. setting professional and life goals while keeping up with weekly to-do lists. The rigor and tenacity I’ve developed also set me up for tough problems in the industry. Having navigated a challenging yet diverse course load at Georgia Tech, I’m able to explore multiple interests simultaneously. Problem-solving can come in many different forms, and I’m always excited about the opportunity to tackle new challenges.

What is your favorite Georgia Tech memory?

My favorite Georgia Tech memories come from the many creative ventures during Dr. Dellaert’s Computational Photography course. I filmed sunset time-lapses of the Atlanta skyline from Jackson Street Bridge, captured the lunar eclipse on my iPhone, recreated panoramas from past photographs, and converted my room into a real-life pinhole camera.