I am an undergrad studying CS with a focus on ML & distributed systems at Stanford.
I'm interested in solving problems at scale involving search, compute, data, systems, and ML. This past winter, I spent a quarter at Oxford as a visiting student at Magdalen College, Oxford; I studied machine learning theory, reinforcement learning, and alignment. This past summer, I worked on lowering the cost of pre-training foundation models at AWS AI on the Deep Engine Science team. At Stanford, I work with Jared Quincy Davis on Foundry as an intern member of the Technical Staff (also affiliated with the Future Data Systems Group). I also work on AI/data at Lux Capital.
MLXserver: a Python endpoint for downloading and performing inference with open-source models optimized for Apple metal ⚙️
Anthropic-Bedrock: SDKs for Anthropic's models on AWS Bedrock (without the annoying auth)
Monosemanticity: a rapid indexing and visualization layer to understand how groups of neurons (features) activate based on the latest mechanistic interpretability research.
Augment: open-source data infrastructure for labeling, RLHF, and augmenting ML training data.
Cambrian: the co-pilot for AI research: ML-powered search, bookmarking, and QA.
Cardinal Compass: Stanford's first natural language course search. Went viral and was featured in the school paper.
CardOptimize: Schedule optimizer for requirements.
Plato: AI pair-programming platform powered by your voice.
Universal Gradient Descent: a smorgasbord on the future of innovation, uniformity of thinking, globalization, surveillance, and what it means to be human.
The AI Arms Race: technical nation-state creation in the age of AI