I'm a product manager with a computer science degree, which means I can write the PRD and then go implement it. I've shipped a multi-tenant inventory platform, an on-device privacy tool, and an NLP screening system that cut manual work by 60%. Right now I'm at NYU finishing a master's and building two products in parallel.
Four systems I built and one experiment I designed. Each started as a written problem statement and ended as something with numbers attached. The figures are drawn from the real repositories and the real design doc, not decoration.
A production-grade, multi-tenant inventory management system — designed as a real product, not a portfolio piece.
An A/B test I designed for Duolingo's onboarding — built to answer a question the company's own most-cited experiment left open.
A privacy firewall for AI tools — it redacts sensitive data from your clipboard before it ever reaches ChatGPT, Claude, or Gemini.
Led a four-person team as Product Lead to cut manual resume screening by 60% with an NLP pipeline.
Turned scattered insurance operations data into a self-serve dashboard built on an 8-KPI framework.
Computer science, then strategy, then product. In that order, on purpose.
| Year | What | Note |
|---|---|---|
| 2025 — | MS Project Management (STEM), NYU | GPA 3.91. Building NexusIMS and SafePaste alongside coursework. |
| 2024 – 25 | Business Strategy Analyst, Faclon Labs | Funded IoT startup. Owned a supply-chain product end to end; +20% reliability. |
| 2024 – 25 | PG Diploma, Data Science — SDBI | CGPA 9.05. ML, NLP, predictive modelling. |
| 2023 – 24 | Product Lead, AI Placement System | Four people, three sprints, twelve UAT cases. First time owning a product. |
| 2021 – 24 | BS Computer Science, KC College Mumbai | CGPA 9.73. The reason I can read the codebase I am speccing. |
I started in computer science and spent three years learning how systems actually fail. Then I found out the expensive failures are almost never technical. They're a team shipping the wrong thing quickly, or arguing for six weeks because nobody wrote down what "churn" means.
So my process is boring on purpose. Write the problem down first. At Faclon Labs that meant sitting with the supply-chain team until the actual constraint surfaced, which was reconciliation lag, not forecasting. On the insurance dashboard it meant refusing to build a chart until eight metrics had agreed formulas. That argument was the deliverable; the dashboard was just the artifact.
Then sequence honestly. RICE is not a magic ranking, it's a way to make the cut defensible when scope pressure arrives mid-sprint — and it always does. On the placement system it's why we shipped the parser and the ranker before anyone's favourite analytics view.
Then build it. This is the part I'd rather not outsource. NexusIMS is 165 source files I wrote against a system design I wrote, and speccing a module marketplace teaches you things about permission scoping that no amount of roadmap review does. I don't think every PM needs to code. I do think the ones who can write better specs.
I'm looking for product, business analyst, and data PM roles starting January 2027, and I'd rather start the conversation early than late. If you have a problem that isn't well defined yet, that's the interesting kind.