I work where AI, product and business meet: scoping a problem with stakeholders, building the solution, and explaining the results to the people who decide.
- Product Management Intern at OneMetric (May - Jul 2026): built an OCR and image-recognition eKYC tool that cut onboarding delays by 70%, and owned an AI/ML intent analyzer for B2B sales that cut lead qualification time by 60%. Two solutions reached production.
- Currently building AI agents and automating workflows end to end, from requirements to deployed tools. Latest: PolicyPilot, a compliance copilot with cited RAG answers and a tool-calling review agent.
- Leadership: Overall Coordinator of the Placement Cell at IIT Jodhpur, and former Overall Coordinator of The Product Club.
PolicyPilot Flagship AI projectAn AI compliance copilot for compliance, operations and legal teams. It answers questions from policy PDFs only from the uploaded text, with page-level citations, and says so when the documents don't cover a question. A tool-calling contract review agent checks a contract clause by clause against those policies and produces an auditable checklist, with guardrails that reject citations to passages it never retrieved. Switch between OpenAI, Claude and Gemini per request; every LLM call is logged for tokens, latency and cost on an admin dashboard. Ships with a 47-question retrieval evaluation and CI for backend and frontend.
|
|
Full-stack digital wallet with OCR-based identity verification, atomic wallet-to-wallet transfers, a SHA-256 hash-chained ledger that makes tampering detectable, and HMAC-signed QR payments. Includes architecture, data model and OWASP-based security review docs, plus a pytest suite.
|
Walk-in interview platform that alerts nearby candidates about hiring drives in time. Geospatial targeting based on each candidate's own travel radius, queued notifications over Telegram, and a FastAPI ML sidecar for speech-to-text and multilingual embeddings.
|
|
Normalized (3NF) SQL Server database of 8 tables for a food and grocery delivery platform, loaded with about 164,000 realistic rows from a Python generator. 20 analysis queries using joins, CTEs and window functions, 9 views, 4 stored procedures, and indexes benchmarked at 45-99% fewer logical reads. One command builds it all and exports 6 KPIs for a Power BI dashboard.
|
Asynchronous data pipeline over the Reddit API combined with 10,000+ tweets. VADER sentiment scoring classifies posts into 3 classes, and daily sentiment is correlated with Bitcoin returns.
|
Languages
AI/ML and Analytics
Backend and Data
Security and FinTech
Product and Consulting
Tools
Product Management Intern, OneMetric | Gurgaon | May 2026 - Jul 2026
- Built an eKYC identity-verification tool with OCR and image recognition, cutting onboarding delays by 70%.
- Owned an AI/ML intent analyzer for B2B sales from scoping to deployment, cutting lead qualification time by 60%.
- Translated stakeholder requirements into user flows and ranked features by value versus effort.
- Validated proofs of concept with users and presented analytics and model explanations to leadership; 2 solutions reached production.
- Overall Coordinator, Placement Cell, IIT Jodhpur (present)
- Overall Coordinator, The Product Club, IIT Jodhpur (Aug 2025 - Apr 2026)
- Assistant Head, E-Cell, Institute Innovation Council (Jul 2024 - May 2025)
- 3rd place, Product Club's Pitch Rush 2025, with a product deck on BluSmart
