Turning raw data
into clear decisions.
Data Analyst — SQL, Python, Power BI. I design the data model, run the analysis, and turn it into a decision someone can act on.
Latest updates
Pathsy — the data model & analytics behind a career-guidance platform
Millions of Indian students finish Class 10 and 12 each year with no structured way to compare what comes next. I structured the decision as data: a 12-table PostgreSQL schema — courses, specializations, career roles, entrance exams via junction tables, 4-stage salary bands — with 78 active courses curated from official board and exam-authority sources. Every relationship is queryable; every recommendation is explainable.
On top sits a transparent weighted-scoring quiz engine: 9 questions carrying per-path JSONB weights — no black box. I validated its calibration in Python with simulated student archetypes, found a real bias toward academic tracks, and redesigned the scoring so high-stakes paths need an explicit commitment signal — 9/9 archetypes now land in the top 3.
The Iran Shock: How an Oil Crisis Reaches Indian Household Budgets
Bottom income quintile carries 2.6× the relative fuel burden of the top. Built a crude-to-pump price transmission model with live rupee data.
104,300 Premature Deaths: India's PM2.5 Air Pollution Burden
Closing the NAAQS–WHO gap could prevent ~27,000 additional deaths per year. Transport and industry are the primary levers.
U.S. Hospital Price Audit: 2.35× Above Global Benchmarks
Federal reference pricing reform projected to save ~$180B annually. 44% of hospitals non-compliant with price transparency rules.
Start from the decision the analysis has to serve — not from the dataset.
Source it or model it, and document exactly where every number came from.
Normalize, deduplicate, and state every assumption out loud.
SQL first; Python where SQL runs out.
One chart per claim. No decoration.
End with what to do next, not just what happened.
- Shipped production features in React/Next.js on PostgreSQL — owned the data model, the queries, and Docker-based deployment end to end.
- Diagnosed data-consistency bugs between application state and the database; the same discipline now drives how I validate analysis inputs before trusting them.
- Owned data accuracy across university web content and produced recurring reports for the web team and department stakeholders.
- Ran Mailchimp campaign workflows end to end — audience segmentation, scheduling, and open/click-rate performance reporting.
- Researched and structured departmental content — turned unstructured academic material into organized, publishable web pages.
- Supported data collection and content accuracy for university web projects.