Product manager who connects the dots. A multidimensional problem solver. Strategy, data, AI and a bias to build, across biopharma analytics, a sports-tech startup and SaaS.
Open to APM / PM / AI PM roles · India or remote. Based in Bangalore, India.
About
I’m a product manager and founder who is happiest when the problem is messy and the goal keeps changing.
I took Hometurf.ai from zero to 4,500+ users across two cities, from tournament management and bookings to AI cameras that cut highlight reels on their own.
Before that I mined 20+ markets for global pharma strategy at IQVIA, and today I work on an AI-integrated ERP with e-commerce merchants.
Different industries, different goals, same habit: find what actually matters, decide, and ship it.
Domains: SaaS & ERP, AI products, Biopharma, Sports tech, E-commerce.
4,500+ users on Hometurf; 10+ integrations shipped; 60% ops effort automated; 70% less lead-gen effort; 20+ markets analysed; 4+ years building.
Range
Strategy
Sizing markets, sharpening positioning, choosing what not to build. 20+ markets analysed at IQVIA · MBA in Marketing.
Skills: Roadmapping, Go-to-Market, Market Research, User Research.
Build
From PRD to a shipped app on Android, iOS and Web. 4,500+ users · 10+ integrations.
Founder responsible for everything: product, technology, business and strategy, from user interviews and PRDs to sprint planning, QA and iterative releases.
Grew the product from bookings into tournament management, then into AI cameras that auto-generate match highlights.
Shipped a cross-platform app (Android, iOS, Web), prototyping fast in Figma, v0 and Gemini Studio.
Built and ran 10+ integrations: WhatsApp, Razorpay, OpenAI, Gemini, YouTube.
Automated internal ops with AI agents and scheduled jobs, cutting repetitive effort by ~60%.
Fulfil.io: Associate Product Consultant
Supply chain & ERP. May 2026 to Present.
The bridge between engineering and e-commerce merchants.
Own merchant integration issues end to end on an AI-integrated ERP, from discovery to resolution.
Find root causes across connected ERP modules and carrier, 3PL and marketplace integrations.
Run continuous discovery: spot recurring merchant pain points, define outcomes, prioritise with Product and Engineering.
Manage the backlog against SLA targets and turn resolutions into repeatable frameworks.
IQVIA: Analyst
Commercial analytics, biopharma. Aug 2022 to Mar 2024.
Data that shaped go-to-market strategy for global pharma.
Cut lead-generation effort by 70% with an automated Python, Excel and LinkedIn Sales Navigator pipeline.
Mined IQVIA’s MIDAS database across 20+ markets and multi-million dollar product portfolios.
Delivered executive-ready dashboards that shaped quarterly commercial strategy.
Caught a recurring data discrepancy in a key client report and added a validation step that prevented downstream errors.
Featured projects
Hometurf.ai
From bookings to tournaments to AI cameras. Zero to 4,500+ users across 2 cities.
. Why I built a territory game for riders on my own, and what the stack looks like.
Biker Battleground is a side project: a mobile-first web app where motorbike and cycling riders claim roads by riding them, hold the crown for a segment, and discover trails where nothing is mapped yet. I started it to scratch my own itch as a rider, and to see how far one person can take a real product, end to end, without a team.
I did the whole thing myself: the game mechanics (how a crown score is calculated from rides, days ridden and speed), the map data (importing OpenStreetMap roads for Bangalore into PostGIS), the product design, and the code. The stack is a React and TypeScript PWA on the front end and Supabase (Postgres, PostGIS, Realtime, Storage) on the back, with about 60 SQL functions doing the actual logic.
The hardest part was not the map or the scoring, it was the social layer: a feed, follows, crews, planned rides and live location sharing, all needing the same care as the core game loop so the app feels alive even before there are many riders on it.
It is a working prototype today, security-reviewed, with a live demo seeded with sample riders. It is not open to real users yet, but you can open the demo and explore the map, feed and leaderboards.
. A small live tool that lets Jev answer questions about a photo, not just text.
imaJev is a small tool I built and shipped live: upload a photo, ask it anything in plain language, or build an if/else decision tree over it, and get back typed, calibrated answers instead of a paragraph of generated text. It came out of using Jev, TypeSafe's System One model, on this portfolio's own AI features and wanting to see what it could do with images, since Jev itself only reads text.
The trick is a three-step pipeline. A vision model (Gemini, called through OpenRouter) describes the photo in exhaustive, literal detail first. A second model call then reshapes your free-text questions, and any if/else conditions you wrote, into Jev's typed primitives: Noul for yes or no, Choice for picking one of a named set, Score for a position on a scale. Jev answers everything in one batched call, and the if/else chain is walked deterministically afterwards, so the outcome is never a guess dressed up as prose.
It is open to anyone, no sign-in, rate-limited per IP so it stays free to run. Images are never written to disk or a database anywhere in the stack. I also built the same three steps as a visual n8n workflow, for anyone who would rather see and edit the pipeline outside the app instead of reading the code.
It is live at imajev.vercel.app if you want to try it.
. A group itinerary editor where an AI drafts the plan and the group changes it through proposals and votes.
TagAlong is a collaborative itinerary editor for group trips. Planning a trip with friends usually means one overworked organizer, a chat thread where decisions get lost, and a loudest-voice-wins plan. I noticed that a shared itinerary is a living document with many contributors who need review and approval, which is exactly the problem version control already solves. So TagAlong borrows GitHub's model: the trip is the repo, the plan is the main branch, a suggested change is a proposal, the group votes, and an approved proposal is locked in as a numbered revision.
The AI does the heavy first draft. Each traveler fills in preference cards that Jev, TypeSafe's judgment model, pre-fills from what is already known, so the quiz takes about a minute. A LangGraph pipeline then plans every day in parallel from real, rated places, checks opening hours, routes and budget in code, and scores how fairly the plan treats each person. The rule behind the design is that the LLM proposes, Jev judges, and code disposes. Votes, quorum, veto and lock-in are always plain SQL, never a model decision.
I wrote the product spec first, a detailed PRD with roles (Captain, Co-pilot, Traveler, Guest), governance presets and success metrics, then built the whole thing solo on Next.js, Supabase with row-level security, LangGraph, OpenRouter and Mapbox. It also imports an existing itinerary from a PDF or photos, and a location-aware assistant answers questions about the trip using read-only SQL that runs as the asking user.
It is live at tagalong-vert.vercel.app. Friends join with a single link, no app and no sign-up.