Tom Qiu

Research Assistant

Yale University

May 2024 - Jan 2025 (9 months)

• Conducted extensive analysis on retrieval-augmented code generation, focusing on dense, sparse, and graph-based RAG techniques, to improve LLM's performance on repo-level generation tasks • Developed an open-source LLM workshop demo featuring a quantized Llama3 model

Vice Project Manager

Yale Alternative Investments

Jan 2024 - Dec 2024 (1 year)

AI Research Assistant @ Gerstein Laboratory

Yale University

Dec 2023 - May 2024 (6 months)

Analyst

Yale Alternative Investments

Sep 2023 - Apr 2024 (8 months)

• Only student organization at Yale partnering with industry-leading PE, VC, HF, & RE funds (acceptance rate ~12%). • Forecasted investment opportunities in green hydrogen for Hannon Armstrong and assisted in managing $9 billion in assets in sustainable infrastructure markets.

Software Engineering Intern

TetraScience

Jun 2022 - Jul 2023 (1 year 2 months)

• Utilized Python and Javascript to convert diversified instrument output files to standardized IDS JSON files, accelerating scientific discovery in the life science industry • Orchestrated a team of high-performing interns, successfully resolving task script and protocol failures across 100 repositories, addressing a wide range of issues including unit and integration testing failures, unresolved package dependencies, formatting errors, linkage issues, and publish and build failures • Spearheaded the development of new data pipelines for various medical instruments, such as the Chemometec NucleoCounter NC-200, with comprehensive unit and integration tests • Streamlined pipeline creation by implementing end-to-end testing leveraging Github

Machine Learning Researcher

The University of Texas at Dallas

May 2022 - Jan 2023 (9 months)

• Applied artificial intelligence to improve drought prediction accuracy, minimizing the adverse impact of drought for communities in endangered areas • Utilized Python to build Long Short Term Memory (LSTM) models and Convolutional Neural Networks (CNN) to analyze conditions leading to drought • Conducted feature ablation to create a standardized dataset containing only essential meteorological and soil data • Provided a benchmark for deep learning models and explained model interpretability Show less

About

Detail

Timeline

Résumé