Portrait of Vedang Asgaonkar

Vedang Asgaonkar

MS Computer Science, Stanford University

vedanga@stanford.edu | GitHub | LinkedIn

I'm a first-year MSCS student at Stanford, specializing in AI. I work with Prof. Jure Leskovec on the Relational Transformer, a foundation model for in-context prediction on relational databases.

Before Stanford I spent two years as a quantitative researcher at Tower Research Capital in London and New York. I studied computer science at IIT Bombay, where I graduated first in my batch and received the President of India Gold Medal for my academic achievements. There I did research on feature acquisition, time-series retrieval and evolutionary game theory.

Publications

Generator Assisted Mixture of Experts for Feature Acquisition in Batch

blog | AAAI'24 | arXiv

Experience

Tower Research Capital

Early member of the mid-frequency trading desk. Built a model architecture for cross-ticker dependencies that composes attention with recurrent time-series layers, and an agentic model-training pipeline that handles 100k+ features while guarding against overfitting and adapting to changes in market regime.

Optiver

Built the market-data input component for delta-one trading on TMX in modern C++. It arbitrates packets across multiple streams to construct the limit order book and parse trades, with cross-stream consistency encoded as a state machine. Received a full-time return offer as a low-latency developer.

Education

Stanford University

Expected June 2028. Coursework in architectures and foundation models.

Indian Institute of Technology Bombay

CGPA 9.97/10, rank 1 in the batch. Advanced proficiency in 13 courses, including Advanced Machine Learning and Advanced Probability.

Awards

President of India Gold Medal

Dilip R. Limaye Academic Excellence Award

JEE Advanced, All India Rank 7

KVPY, All India Rank 1

INPhO and INChO national camps