LangGraph multi-agent text-to-SQL system for querying global health KPI data in natural language.
Reduced repeated response latency from 2-5 minutes to seconds while improving reliability for KPI questions. In development View projectAbout Me
I am a Harvard biomedical engineering student pursuing a secondary in computer science and competing in NCAA Division I track and cross country. My work sits where biological context meets technical execution: protein language models, public health AI agents, epidemiology, RNA-seq analysis, wearable physiology, and geospatial tools that turn complex data into useful systems.
Projects
Browse selected work across Python, Pandas, NumPy, Machine Learning, Bioinformatics, R, SQL, LangGraph / LangChain, and more.
LangGraph multi-agent text-to-SQL system for querying global health KPI data in natural language.
Reduced repeated response latency from 2-5 minutes to seconds while improving reliability for KPI questions. In development View projectEnd-to-end RNA-seq and GSEA workflow for whole-blood samples exposed to temperature stress.
Validated enrichment scores against GSEA Software and identified heat shock and proteostasis pathway enrichment. Completed View projectEnsemble and Bayesian models estimating blood lactate from wearable physiological signals.
Improved test R2 from 0.40 with baseline linear regression to 0.94. Completed View projectKalshi NBA prediction-market analysis joined to ESPN win-probability shocks to test whether markets overreact or underreact in game.
Found that true one-minute overreactions were rare, while sequence models tracked immediate market responses substantially better than tabular baselines. Completed View projectFlask and Leaflet.js web app for exploring Strava activities as interactive routes with map themes, geosearch, and activity details.
Developed a web interface for viewing past Strava activities with route details, activity statistics, customizable map themes, and location search. Completed View projectBioinformatics pipeline for identifying SNPs associated with canine degenerative myelopathy from low-coverage whole-genome sequencing data.
Created a documented pipeline sequence for moving from raw canine whole-genome reads to SNP and gene investigation. Completed View projectPython playlist converter that scrapes dynamically generated Musi playlist URLs and recreates them as Spotify playlists.
Created a converter for moving playlist content from Musi iOS playlist URLs into Spotify. Completed View projectExperience
Fine-tuning PPLM-Affinity with PyTorch and a CUDA/SLURM training pipeline to predict SARS-CoV-2 spike binding affinities to neutralizing antibodies and ACE2.
Conducted Cox proportional hazards modeling on 400,000+ Japanese adults to identify individual and neighborhood socioeconomic predictors of obesity onset.
Led Basecamp-based project systems and maintain the public website for a national consortium of 100+ climate and health researchers.
Managed reproducible RNAi experimental pipelines and quantitative phenotypic data across 200+ planarians.
Supported Breast Cancer Center operations by assisting 10+ patients daily, resolving Epic EHR workflow issues, and supporting rehabilitation staff during patient evaluation.
Lead a 50+ member organization delivering technology-driven consulting and research projects for global health partners, including a charity 5k that has raised $10,000+ for cancer research and care.
Check out TGHI eventsExplore upcoming TGHI events for student-led conversations, workshops, and programs at the intersection of technology and global health.