I'm currently a student at Stanford studying math and computer science. My interests lie in artificial intelligence research, especially areas related to vision and decision making. I care about the people around me, contributing to society, and enjoying the journey.
Phone: (703) 939-0540
Theory: Algorithms (CS 161, 168), Data Structures (CS 166), Complexity/Computability Theory (CS 154)
Systems: Systems I/II (CS 107, 110), OS Theory (CS 140), Compilers (CS 143), Parallel (149), Security (CS 155, 255), Databases (CS 245)
AI: NLP (CS 224n), Graphical Models (CS 228), Reinforcement Learning (CS 234, 238), Generative Models (CS 236)
Places I've been, things I've done. Startups, big tech, research, and sports.
Improving autonomous navigation in urban environments with 3D information. Part of the JackRabbot team.
I'll be joining the Global Fixed Income team as a Quantitative Researcher. Stay tuned to learn more!
2nd engineer at Sequoia-backed startup building E2EE collaboration tools. Built filesystem and designed security model.
Researched and wrote on the Australian economy, focusing on pandemic fiscal policy impacts and household debt.
Built infrastructure to create topic stories. Created ML models to determine quality and intelligently rank stories content.
Identified poor content on Capital One’s learning platform with an AWS serverless pipeline, reducing bad content by 30%.
Worked as senior referee at GF Hoops basketball leagues, reffing elementary, middle, and high schoolers.
Streamlined sales prospecting by NLP-based task analysis with Elasticsearch and a React front end.
A short list of highlighted projects. Some are research, some for class, and some hobby.
The first large-scale 2D/3D egocentric robotics dataset in human environments. Published in TPAMI.
An online, real-time, 2D-3D tracker for autonomous robots and cars. Published in IROS 2020.
Competed in Battlecode, a month long MIT AI strategy competition. Tied 9th 2021, 2nd 2020, 1st 2019, tied 9th 2018.
Explored training a family of autoregressive models for improved image completion and sublinear image generation.
Adapted ideas from Brodal queues and skew binomial heaps for improved item deletion times in a purely functional setting.
A computer vision system to calculate limb size from phone images. Used to track the state of Lymphedema swelling.
Automated identification of ISIS propaganda accounts on social media. Presented at NSA and Raytheon.