RAVE is a research group in UVA's School of Data Science run by PI Tom Hartvigsen. We work on responsible AI in constantly-changing environments, especially those that exist in healthcare environments and regularly publish our research in the top ML/NLP/Medicine venues.

Research Philosophy

• Great research requires deep insights

• Deep insights require deep understanding

• Deep understanding requires building from scratch

• If you can't explain your ideas clearly without notes, you don't understand your ideas well-enough yet

Research Strategy

• Pick problems with few competitors --- ideally, define your own! Do something nobody's ever done!

  • Pick problems we are uniquely qualified to tackle

  • Build your own research area and let people come to you over time

  • Popular areas are exciting and potentially more-impactful, but also more-competitive and so much higher risk

• Ask questions others are incentivized not to ask --- makes your work unique and important, avoid competing with heavily-resources industry labs

• First think long-term and big picture --- choose problems that might become popular 2-4 years from now (and motivate it really well)

  • There's almost never any low-hanging fruit: If it's easy, thousands of people may see the same thing and it becomes a race

  • Avoid reactive research where you chase the ball

• Then think short-term and goal-oriented --- After picking a problem, solve it by any means possible, ignoring other (distracting) fun ideas until you finish

• Prioritize problem-centric innovation --- the tools we build and use always change, but people's needs change slowly. Setting a clear goal lets you assess solution fit and identify problems that require innovation to solve.

• Don't be an "ideas person" who just comes up with many cool ideas. That's not the hard part---Ideas are cheap and easy. Execution is hard and valuable.

• Challenge conventional wisdom: If many people believe something, improving our understanding is high-impact 

Mentoring Philosophy

While we conduct exciting research in a fast-paced field, a healthy work-life balance and general wellbeing are key to sustained research success.

I aim to broaden participation in machine learning research in all aspects. Intersecting more perspectives fosters creativity, so folks from all walks of life are welcome in my group and I do my best to accommodate everyone's situation.

Conducting successful research should be fun and involve diving deep into details. To facilitate this, students and I work together to define concrete problems we are both excited about.

I encourage and foster independence at all stages of students' degrees.

A big component of doing a PhD is self-discovery, so I encourage students to try new things and read broadly.

Supervision logistics: I meet each student 1:1 for 1 hour per week. This meeting is student-led and usually focuses on research updates and feedback. I try to be available as much as possible for paper-writing and other support.

*I don't condone the overuse of "AI" but "RAVE" sounds cooler than "RMLVE"