AI Product Manager with a CS foundation — I bridge LLM capabilities and product execution.
I don't just spec AI products. I build, deploy, and measure them.
- 🤖 Core expertise: LLM Pipelines · RAG Systems · Computer Vision · NLP
- 📐 PM craft: PRD writing · Agile/Scrum · Backlog management · Stakeholder alignment
- ⚽ Built FootIQ — full-stack football analytics platform with CV + edge computing
- 📊 Shipped Groww Pulsator — LLM app review analytics · 6,792 reviews processed · NPS +78
- 🎓 NextLeap PM Fellow | B.Tech CS, Amity University
- 📍 Gurugram, India | Available from July 2026
Open To: AI PM · APM · Product Analyst — early-stage AI startups & mid-size product companies
Languages
Frontend
Backend & Databases
Cloud, DevOps & Tooling
AI / ML Tooling
| Domain | Proficiency | Details |
|---|---|---|
| LLM Pipelines | ████████░░ Advanced |
GPT-4, Claude, Gemini integrations; prompt engineering, structured outputs |
| RAG Systems | ████████░░ Advanced |
Vector DBs, embedding pipelines, document-grounded Q&A |
| NLP | ███████░░░ Proficient |
Sentiment analysis, text classification, app review mining at scale |
| Computer Vision | ██████░░░░ Proficient |
Object detection, player tracking, edge inference deployment |
| AI Product Strategy | █████████░ Expert |
Roadmapping, LLM capability scoping, build-vs-buy decisions |
| Prompt Engineering | ████████░░ Advanced |
System prompts, CoT, few-shot, XML structuring, eval design |
| Agentic Workflows | ██████░░░░ Proficient |
n8n automation, tool-use agents, multi-step LLM orchestration |
📊 Groww Pulsator — LLM-Powered App Review Analytics
AI-native product intelligence tool that ingests, processes, and extracts structured insights from thousands of app store reviews using LLM pipelines — deployed and production-ready.
| Attribute | Detail |
|---|---|
| Stack | Python · LangChain · GPT-4 · Streamlit · PostgreSQL |
| Scale | 6,792 reviews processed end-to-end |
| Performance | NPS of +78 among early users |
| AI Layer | LLM-powered sentiment clustering · feature extraction · theme detection |
| Security | API key isolation · input sanitization · rate-limited pipeline |
| Impact | Cuts manual review analysis from hours to minutes |
| Repository | View on GitHub → |
Groww Pulsator demonstrates the full AI PM loop: product hypothesis → technical build → deployed app → measurable outcome. Built as part of the NextLeap PM Fellowship's Learn in Public Challenge 4, it auto-clusters user feedback into product themes and generates PM-ready summaries that any product team can act on immediately.
⚽ FootIQ — Football Analytics Platform
Full-stack football analytics platform leveraging computer vision and edge computing to deliver real-time player performance intelligence.
| Attribute | Detail |
|---|---|
| Stack | Python · OpenCV · FastAPI · React · Edge Computing |
| Scale | Multi-player real-time tracking pipeline |
| Performance | Low-latency edge inference for live match analysis |
| AI Layer | Object detection · movement tracking · performance heuristics |
| Security | Edge-isolated inference · no raw video egress |
| Impact | End-to-end sports AI: data pipeline → insight dashboard |
| Repository | View on GitHub → |
FootIQ is the flagship project — a full-stack sports intelligence system built from the ground up. Combines computer vision for player tracking with an analytics layer that surfaces match insights. Inspired by production-grade work at WCO Global / Elle Global, it demonstrates AI PM execution at the infrastructure level, not just the spec level.
🏦 Mutual Fund FAQ Chatbot — RAG-Powered Financial Assistant
Production Next.js + FastAPI application powered by a RAG pipeline to answer mutual fund queries with grounded, document-sourced responses — rearchitected under time pressure for free-tier infrastructure.
| Attribute | Detail |
|---|---|
| Stack | Next.js · TypeScript · FastAPI · Groq LLM · Vector DB · Tailwind CSS |
| Architecture | Scraping → chunking → embeddings → retrieval → generation |
| AI Layer | Full RAG pipeline — hallucination-resistant, document-grounded answers |
| Security | PII filters · document-scoped retrieval · no external data leakage |
| Status | Deployed on free-tier infra (Vercel + Render) |
| Repository | View on GitHub → |
Built during the NextLeap PM Fellowship under real infrastructure constraints. The full pipeline — scraping, chunking, embeddings, LLM integration, and PII filtering — was rearchitected mid-build for free-tier deployment. Demonstrates LLM product execution under pressure.
🎮 GameSense AI Copilot — In-Game Intelligence Layer
AI copilot for real-time game strategy assistance — full PRD from market sizing through GTM roadmap and monetization model.
| Attribute | Detail |
|---|---|
| Stack | React · TypeScript · RAG + LLM Architecture |
| Market | $4.7B market sizing · 3 JTBD personas |
| PM Artifacts | Full PRD · User Stories with Acceptance Criteria · Risk Matrix · GTM Roadmap |
| AI Layer | Real-time LLM strategy suggestions · context-aware guidance |
| PM Depth | MoSCoW prioritization · 4-phase GTM · monetization model |
| Repository | View on GitHub → |
A PM-led product build with a $4.7B TAM analysis, 3 validated JTBD personas, and a 4-phase GTM roadmap with monetization model. Demonstrates the full PM lifecycle from market validation through production-ready spec.
🍽️ Zomato AI Recommender + 3 More AI Products
Four AI digital products shipped end-to-end: user research, product vision, technical flows, and measurable outcomes.
| Product | Stack | Key Result |
|---|---|---|
| Zomato AI Recommender | Streamlit · GPT API · Python | Personalized restaurant discovery with LLM reasoning |
| HealthGuard | FastAPI · ViT · Python | Visual health monitoring with Vision Transformer |
| TalentScout | FastAPI · GPT API · Python | AI-powered talent matching pipeline |
| Trend Predictor | Prophet · Python · Streamlit | 12% forecast accuracy improvement |
| Attribute | Detail |
|---|---|
| Aggregate Impact | 30% latency reduction across pipelines · 12% forecast accuracy gain |
| PM Depth | User research · product vision · flow definition for each product |
| Repository | View on GitHub → |
Four AI products shipped with consistent PM rigor: each started with user research and product vision before a single line of code. Demonstrates breadth across recommendation, health tech, talent, and forecasting verticals.
Football Analytics · Aug 2025 – May 2026
Led end-to-end product lifecycle for a multi-camera football analytics system — from PRD authoring through edge hardware deployment.
- Authored PRDs and defined specs for RTSP streaming and AV sync across multi-camera pipeline
- Coordinated hardware and engineering teams to ship on Raspberry Pi 5 edge devices
- Identified latency bottlenecks via data analysis and drove engineering fixes improving system reliability
- Managed stakeholder alignment across vendors and internal teams throughout the product lifecycle
Fintech · May – Jun 2025
Mapped end-to-end customer journeys and drove API prioritization for IPO data flows on a digital fintech platform.
- Mapped customer journey for IPO data flows; prioritized API requirements with stakeholders delivering 40% faster response times
- Built financial insight dashboards using PostgreSQL and Matplotlib for end-user analytics
- Collaborated with engineering to translate user needs into measurable product improvements
Early-Stage Startup · Jan – Feb 2025
Managed a 6-member cross-functional team in a 0→1 environment with full Agile/Scrum ownership.
- Managed 6-member cross-functional team via Agile/Scrum — daily standups, sprint planning, retrospectives
- Owned JIRA backlog end-to-end; achieved 100% on-time delivery across all sprint commitments
- Aligned team output with stakeholder expectations through structured OKR tracking
AI / Rural Tech · Sep – Dec 2024
Scoped and shipped an AI chatbot for rural communities — from customer discovery through QA and production delivery.
- Led customer discovery to identify high-value use cases: agriculture schemes, subsidies, and rural information access
- Defined feature set, managed design-to-QA delivery for a fine-tuned DialoGPT agriculture chatbot
- Established perplexity-based success metrics to measure LLM output quality in production
| Recognition | Details |
|---|---|
| 🎓 NextLeap PM Fellow | Completed structured PM Fellowship — Learn in Public challenge series |
| 📊 Groww Pulsator | 6,792 reviews processed · NPS +78 · Deployed production LLM app |
| ⚽ FootIQ — Flagship | End-to-end sports AI platform: computer vision + backend + dashboard |
| ⚡ 40% API Speedup | Delivered at Bluestock Fintech via customer journey mapping & API prioritization |
| 🎯 100% On-Time Delivery | Led 6-member team at Excelerate across all sprint commitments |
| 📝 PM Artifact Library | 5+ full PRDs · 3+ wireframe kits · $4.7B market sizing · 1 PM Handbook |
| 🔍 AI Product Teardowns | VocalLabs.ai · BlinkMoney · Reddit · Make.com — published publicly |
Education
Fellowship
Courses & Bootcamps
currently:
learning:
- "Advanced RAG architectures and LLM evaluation frameworks"
- "Agentic workflow design with LangGraph and n8n"
- "AI product metrics — measuring LLM output quality at scale"
building:
- "FootIQ v2 — expanded analytics pipeline and live dashboard"
- "AI PM portfolio with deployed, measurable products"
exploring:
- "Multi-modal LLM product applications"
- "Edge AI for sports and fitness tech"
- "AI-native B2B SaaS patterns and pricing models"
open_to:
roles:
- "AI Product Manager (APM)"
- "Product Analyst"
companies:
- "Early-stage AI startups"
- "Mid-size product companies"
location: "Gurugram, India | Open to Remote"
available_from: "June 2026"
"The best AI PMs are the ones who can build — and the best builders are the ones who think like PMs."
