Developing next-generation graph neural network attention mechanisms. Achieved SoTA on inductive link prediction. Now training a genomics foundation model on 1,024 H100s. It'll probably be the largest bio ML model released to the research community.

Contrary is a talent and research-driven investment firm. We back a small number of exceptional people and the companies they build, from seed to scale.

Led the development of Baymax, the autonomous helping hand for the disabled and the visually impaired. We took 1st place among 1,124 university competitors and scored $10,000 + round-trip tickets to Tokyo.

I developed a novel matching algorithm to equitably assign 50,000+ new students to San Francisco public elementary schools from 2024 to 2034 by predicting their future test scores, family education level, household income, and approximate commute to school from geographic and historical data. My algorithmic design, dubbed the Equity Tiebreaker, was adopted by the San Francisco Board of Education in Board Policy 5101.2 to maximize educational opportunities for the city's low-income, African American, Latinx, and Pacific Islander youth.I was invited to present my findings to faculty at the Stanford School of Engineering, showcase my work to 300+ prospective Stanford students & their families at the 2023 Stanford April Symposium of Undergraduate Research & Public Service, and I continue to advise San Francisco Unified School District executives & staff at the superintendent’s office on algorithmic approaches to resource allocation in public education.

I organized and directed the largest global study on student learning with virtual reality, featuring 160 high school participants from 12 countries spanning Casablanca, Morocco to Antibes, France to Anaheim, California.I employed R, Python, and LLMs for the statistical analysis and data visualization of ~22,900 collected multimedia datapoints from focus group recordings, collaborative student-generated content, free-response survey answers, and questionnaires. I coauthored the paper "Advancing Conversational Learning with Learner-Created Virtual Environmental Field Trips" and was honored as the primary collaborator in my labmate's PhD dissertation defense.

I connected 1,802 high school students with internship opportunities at nonprofits worldwide, helping raise $98,727 for charitable causes and educate 503,604 underserved youth. I managed our international team of 64 staff; directed social media marketing campaigns to build a community of 2,800 international youth volunteers (across our Discord, WhatsApp, and email list); and struck partnerships with 45 multinational nonprofits.Examples of our work include preserving 400 acres of rainforest in the Congo Basin via fundraising with Endangered Species International; developing STEM curriculum for Indian students in collaboration with MakerGhat; and distributing mental health resources to 28 high schools in California with Boldly Me.

I directed the engineering of 110lb, autonomous robots with a team of 31 students to become internationally recognized by the FIRST Robotics Competition (FRC) for innovative code & design (FRC Industrial Design Award) and take second place at the FRC Silicon Valley Regional as the youngest team at the event among 2 former world champions.I managed robot design, mechanical build, electrical engineering, and software integration; developed computer vision pipelines for autonomous navigation; and raised & allocated $60,000+ in sponsorships from donors & organizations including Apple, NVIDIA, and NASA.