• Led a group of 5 postdocs and PhD students to develop large-scale scientific workflows and deep learning recommendation systems for polymer materials on the cloud with Docker, AWS S3, Lambda, ECS, and RDS.

• 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.

• Greatly improved rendering (LCP) of large forms (80+ form fields) from 4.12s to 1.79s (230%faster) by implementing an SPA approach for the Google’s Chrome Status Platform web app which has 130k+ monthly visitors.

• 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.

• Developed high performance computing workflows and conducted data-driven research for battery design using simulation and machine learning methods (6 peer-reviewed papers, 4 as first author)

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