Data Scientist / ML Engineer— New York[ available full-time · january 2027 ]
I build production ML and agentic systems — pipelines, evals, and answers that cite their sources.
Data science intern at Novo Nordisk, working on the rare-disease and Wegovy/Ozempic portfolios. M.S. in Data Science at Columbia. Before that, three years at ZS building ML for healthcare clients.
Five specialized agents that plan, write, validate, chart, and narrate SQL analytics — every number cited back to source rows.
5 agents on a typed message bus·×3 retry budget, exponential backoff·100% numbers cited to source rows
FastAPI · LangChain · DuckDB / Postgres / Snowflake · OpenAI function calling · matplotlib · pytest
A production model that flags likely BTC patients ~45 days before claims data confirms them — scoring 250M patient-claims every month.
~45d earlier identification·250M patient-claims scored / month·PMSA '25 methodology presented
PySpark · XGBoost · SHAP · K-means / GMM · NLP clustering · MLflow
A hybrid rules + LLM + retrieval pipeline that standardized 7M+ farmer transactions into the credit signal behind microloans.
99% classification accuracy·−95% manual-review volume·7M+ transactions standardized
Python · RAG · LLM classification · REST
SEC-filing Q&A with line-level citations — multi-stage retrieval over ChromaDB, evaluated with Claude as judge, live on Cloud Run.
hrs→s time-to-insight·100% answers grounded in retrieved context·3 eval dimensions, LLM-as-judge
FastAPI · ChromaDB · LangChain · text-embedding-3-large · Streamlit · GCP Cloud Run
GAFFER — a team-strength model for the 2026 World Cup: Elo + a Dixon-Coles goal model + Transfermarkt squad value, run through 50,000 Monte Carlo simulations for live title odds that update as the games are played.
Live classroom theme extraction — students answer via QR, an LLM clusters responses into themed cards every 10s over SSE, with a 5-model fallback chain.
Conversations with 60+ historical figures under era-appropriate knowledge boundaries; Dinner-Party mode runs 2–5 figure multi-agent dialogue.
Narrow-scope chatbot that flags APA 7 / MLA 9 / Chicago 17 violations with rule-IDs and quoted evidence; three-method eval suite, 30+ test cases.
Chrome extension that drafts X replies in three rhetorical angles — image-aware context extraction, voice learning from selections, streaming output.
New York · Summer 2026
currentPune · Feb 2022 — Jun 2025
Promoted to Associate Consultant in 4 cycles (typical: 5). Expert Associate and Insight Illuminator awards.
New York · 2025 — present
I work on the part of machine learning that starts after the demo.
Pipelines, evals, drift, citations — the work that keeps a model trustworthy after it ships. That’s where I’ve spent most of my time. Right now I’m building agents that plan, query, and cite their sources.
Columbia University
M.S. Data Science
Aug 2025 — Dec 2026
TA — Business Analytics II (Foundations of AI) · Hollywood & Big Data, Columbia Business School
Vellore Institute of Technology
B.Tech, Electronics & Communication Engineering
Jul 2018 — May 2022
Special Achiever Award · Merit Scholarship