Product manager who builds. I take AI products from the first PRD to a live pilot, and I instrument them so the metrics say whether they worked.

Currently finishing an MS in Business Analytics at UC San Diego (Rady) and leading product on Thrive, 3 AI products for 183 MSBA students.

User analytics for GenAI assistants, built as an end-to-end PM case study on 3.2 million real conversations. It answers four questions a PM who owns an assistant usually can't: who uses it, how often, for what, and where it fails. It ships a browser-side dashboard (DuckDB-WASM, no server) and the full product write-up: PRD, metrics framework, trends report, and retro.

Live site · Dashboard · Case study · Code

• UC San Diego, Rady: launched Thrive, 3 AI products replacing 5 student tools for 183 MSBA students, live in production within 7 weeks with a 4-person team. Shipped a cited-only chatbot with guardrails validated on a 94-question eval bench, then raised its accuracy from 17/27 to 25/27 known-answer questions. The case to retire 3 paid tools projects ~$64K/yr saved.
• Kevali, first product hire: recovered 21% of abandoned carts and grew conversion 15% on a $186K-a-quarter D2C storefront.
• Tap Invest: launched a recommender that lifted engagement 31% across $12M AUM, and cut underwriting from 3 days to 1.
• BCG: built the route-optimization and demand-forecasting models behind a $50M ride-sharing client's plan to electrify half its fleet.

Product strategy · roadmapping · PRDs · experimentation & A/B testing · Python · SQL · DuckDB · RAG & agentic systems · evaluation design

🌐 shankard.com · 💼 LinkedIn · 📄 Resume · ✉️ shankar.deenadayalan@outlook.com