Data Scientist — LLM & multi-agent systems, end-to-end.
about
I graduated from George Washington University with a Master's in Data Science and currently work as a Data Science Research Assistant at the Regulatory Studies Center.
In this role, I lead RegStats and develop data-driven dashboards in Python, building on my previous experience as a Software Engineer in Data at Vikas Constructions.
From deep technical execution to explaining the final product to non-technical audiences, I own the process from start to finish.
selected work
A multi-agent system that detects cross-jurisdiction compliance conflicts across SEC, FCA, and SFDR/EU Taxonomy — mapping 300+ regulatory rules into typed schemas, with LangGraph routing on LLaMA, FastAPI, Postgres, and Pinecone retrieval, Dockerized on Railway behind a React UI that returns rule-level verdicts with citations.
Real-time credit-card fraud detection running XGBoost and Isolation Forest as parallel scoring agents, with a LLaMA 3.3 70B layer generating human-readable explanations for each flagged transaction. Full pipeline deployed on Railway with GitHub CI/CD at sub-200ms end-to-end latency.
A Python dashboard (Streamlit, Plotly, pandas) that automates collection of crude oil, natural gas, and product fundamentals from the EIA API, refreshing weekly supply, demand, and price data. Branded, interactive visualizations — including a five-year seasonal range model for crude and gas inventories — benchmark stocks against historical norms to flag market deviations at a glance.
experience
Own and lead RegStats, the centre's open-source pipeline collecting 15+ federal rulemaking data series (Federal Register, CRA, CFR, Unified Agenda) into 5 dashboards; built an automated validation layer that blocks malformed data; 130+ commits to the public codebase; published quantitative analysis of federal rulemaking trends.
Ran discovery across 15 projects to map procurement and inventory workflows, shipping KPI dashboards that cut material wastage ~23%; built ETL pipelines consolidating 3+ disconnected sources into a unified SQL model; structured raw cost data into GST- and RERA-compliant filing formats.
Led weekly labs and office hours for 20 graduate students (data wrangling, ggplot2, statistical modeling); as AI & ML Lead, mentored 100+ students and secured $1,000 in bootcamp sponsorship.
education
Global Leaders Award ($17,100 scholarship).
skills