Built suite of Python packages - PMD and PSP (35k+ total downloads) to automate simulation data generation and analysis and enable high-throughput virtual screening for materials research.

Built full-stack web dashboard application for data visualization and real-time inference prediction with React, TypeScript, NodeJS, FastAPI, TailwindCSS, PostgreSQL to allow 1k+ daily users to query machine learning predictions.

Augmented experimental datasets with simulation data and applied multi-task neural network models to improve average R2 of multiple models from 0.6 to 0.9.

Simplified Flask REST API protocols (32% less lines of code) for improved maintainability and scalability.

Boosted the unit test coverage (average over 80%) of code for both frontend components and backend API routes.

Led weekly discussion as a TA on scientific topics and guided 100+ students for scientific programming.