• Writing custom visualization tools using Pandas/SQL to analyze market microstructure• Developed efficient min-max optimization method in SQL to interpolate missing trades from intermittent market updates, match and compare with historical data, and label with predicted trade side• Predicting post-trade order flow imbalance using regression model on pre-trade price dislocation, exponential moving average volume, and other features from historical data• Implementing custom Python functions to vectorize computations that are not supported by Pandas or Numpy• Optimizing algorithmic trading strategy based on predictive model

• Studying open conjecture in combinatorics involving x-rays of permutation matrices and lattice paths• Designed algorithm to efficiently sample steps in a Markov chain on a class of lattice paths

Simulating strategies for modified Prisoner’s Dilemma (and similar problems) at MIT CS-AI Lab

• Developed algorithms to classify user questions for Inkeep’s technical search and chatbot• Devised tree-based clustering model, allowing for hierarchical clustering with adaptable granularity levels• Packaged all developed models and training pipeline into a library for seamless integration

Writing constant propagation compiler logic for Bedrock2 programming language at MIT CS-AI Lab and mathematically proving algorithm using Coq Proof Assistant.