AI Product Manager — I build LLM products end to end: RAG, multi‑agent systems, evals, and guardrails.
4+ years shipping enterprise B2B SaaS · Duke MEM '25.
Focus
RAG
4+ years shipping enterprise B2B SaaS · Duke MEM '25.
More about me
I'm a product manager who likes the messy early part — taking an ambiguous problem, finding the real user need underneath it, and shipping the thing that solves it. Over 4+ years I've delivered $520K+ in cost savings on a platform serving 80M+ people, launched a feature to 10K users, and picked up a Duke MEM along the way. Now I build AI products end to end — currently Tear, a multi‑agent research platform that cuts product research from 4 hours to about 10 minutes.
Education
Duke University
Master of Engineering Management '25
Undergrad
B.Tech Bioinformatics
Experience
4+ years
Building from the ground up
Multi-agent pipeline: model routing, retrieval, evals and guardrails built in from day one.
4 hours → ~10 minutes.
Impact
$520K
Annual savings
Infosys · APM
80M+
Users served
Infosys · APM
300M+
Records managed
Infosys · APM
16%
DAU lift
Sparc · AI PM
10K
Users at launch
Sparc · AI PM
64%
Research time cut
Sparc · Intern
Work
Sparc
Feb 2026 – Present
AI Product Manager
Own the AI product lifecycle end to end; latest launch reached 10K users with a 16% DAU lift.
Sparc
May – Dec 2025
AI Product Intern
Mapped the subscription funnel and shipped an LLM + RAG research bot that cut competitive research time 64%.
Infosys
Dec 2021 – Aug 2024
Associate Product Manager
Owned the roadmap for India's national tax platform — 300M+ records, $520K in annual savings.
My approach
01
Discover
Every product starts with a gap: the space between what users say they want and what they actually need. I spend my time here first — talking to users, mapping how they actually work, and asking why until the real problem shows itself.
02
Define
Not every problem deserves a build. Before anything ships, I pressure-test it on four fronts — is it valuable, usable, feasible, and viable — then turn the strongest opportunity into a clear spec with success metrics attached.
03
Deliver
Ship small, ship often. I work in tight loops with engineering and design — prototype, test, refine — and for AI products, evals and guardrails are part of the build, not an afterthought.
04
Measure
Launch is the midpoint, not the finish. Every release gets a number attached — adoption, engagement, revenue — and the data decides what happens next.
Projects
Warner Bros. — Can they rewrite the script on entertainment?
ProblemDrowning in debt while its biggest franchises lose relevance and streaming rivals eat its lunch.
My callRevive iconic IPs with fan collaboration, unlock new revenue through gaming and VR, and use AI to cut production costs — while making Max streaming actually profitable.
So whatA clear path to position Warner Bros. as an entertainment ecosystem, not just a movie studio.
Tesla — Competitive strategy stress‑test
ProblemTesla's lead is eroding as legacy automakers and Chinese EV makers close the gap on price and range.
My callStress-tested its vertical-integration and supply-chain strategy against demand and pricing shocks — defend the moats worth the capital, open up the rest.
So whatA framework for deciding which advantages compound — and which quietly become liabilities.
Colorectal cancer research
Utilized Next-Generation Sequencing to analyze genomic data and pinpoint hereditary genetic mutations linked to familial colorectal cancer. This research aims to improve targeted diagnostics and personalized treatments for high-risk patients.
COVID‑19 mortality data analysis
Applied logistic regression and clustering algorithms to COVID-19 mortality data to evaluate demographic risk factors. Validated media reports by statistically quantifying the increased vulnerability of older adults and women.
Life beyond the resume
The things that don't fit on a resume but explain how I think.
