I ship reliable, human-centered AI inside a Fortune 100 — and the patterns I learn doing it.
From an engine-block factory floor to enterprise AI products. Each chapter taught me something different about what it actually takes to ship.
Product strategy across DMS, TMS, and EU MIR — three regulated life-sciences products on Salesforce Lightning, serving 200+ enterprise life sciences & med device customers. AI-assisted EU MIR with human-in-the-loop, plus other regulatory agents at POC stage.
0-to-1 MES for battery gigafactories. Co-created the architecture with Honeywell Fellows; shipped first alpha in 15 months — against the typical 24+. $100M+ revenue lane, 2 patents filed from my team.
Industrial Gen AI assistant for plant operators. Multi-DB RAG across PostgreSQL, Druid, Neo4j, and Postgres Vector on GPT-4o via Azure AI Foundry — with hallucination guardrails I built before Honeywell had an AI platform. 40% adoption lift, 30% faster decisions.
Solo PM on a connected-worker platform across Android, iOS, Realwear (head-mounted tablet), and Web. Coordinated 30 cross-functional stakeholders. Took adoption from 500 to 2,500 industrial users in 12 months.
Led modernization of 5+ legacy PHP enterprise apps to Azure SaaS for Sempra Energy — 3 scrum teams, 20+ engineers, multi-year strategic account. Ran pre-sales for Product Engineering at an 85% win ratio, contributing 15% to company revenue growth.
Three-month FMCG marketing rotation across rural Maharashtra. 3,000+ retail outlets analyzed in person. Rolled out mobile-based automated ordering at the retailer level — 15% market expansion in the geography.
Two stops at Tata Motors before the Cloud era — where I learned that operations is product management with steel.
Side projects where I stress-test the patterns I write about — agent orchestration, RAG, eval harnesses.
70+ composable skills for AI PM workflows, organized in three layers — Thinking → Judgment → Craft. Distributed as a single Claude plugin with a self-correcting orchestrator.
RAG + knowledge graph + LLM eval harness. A browsable pattern library for AI builders, grounded in real implementations.
100+ deep-dive articles — my working notes from shipping AI products inside a Fortune 100. Published on this site and on Substack.