kishor@bengaluru:~$ ai-product-manager --mode builder
I don't just plan AI products. I build them.ship them.architect them.deploy them.
I'm Kishor Goswami — AI Product Manager, Product Owner and Business Analyst in Bengaluru, with an AI engineering core. I run discovery, write the user stories, design the solution architecture — and then build and deploy it end to end. That's my edge.
01 — The Edge
Two brains. One person.
Most PMs hand off a spec and hope. I close the loop between the boardroom and the codebase.
BRAIN / 01
The Product Mind
Discovery sessions, stakeholder workshops, PRDs, user stories with crisp acceptance criteria, scope-gap analysis, sprint execution, KPI & adoption tracking. I translate messy business reality into a roadmap teams can actually ship.
BRAIN / 02
The Builder's Hands
B.Tech in CS (AI & ML). I design RAG pipelines, orchestrate LangGraph multi-agent systems, wire up FastAPI + React apps, tune prompts, and deploy Agentforce agents into live enterprise orgs. The spec I write is one I can build.
// the difference, as a diff:pm.write_spec(); pm.hand_off(); pm.hope();kishor.write_spec(); kishor.build(); kishor.ship(); // same hands, end to end
02 — Selected Work
Shipped, not slide-ware.
A few things I've architected, built, or product-owned — from enterprise AI transformations to weekend builds that went further than planned.
Leading GCC Government Contractor — AI Tender Management Transformation
Rebuilt an entire tender lifecycle: AI agents that analyze opportunities, decide bid/no-bid, suggest partner vendors, draft communications, and generate proposal documents. Humans now review and approve — AI does the rest.
BA Co-Pilot — Requirement Intelligence Platform
A multi-agent platform that automates the BA workflow I live every day: discovery plans, meeting minutes, scope-gap analysis against SOWs, and export-ready Agile user stories — with actor-critic LangGraph loops for quality.
Leading Indian Housing Developer — Three-Agent AI Suite
A grounded pre-sales chatbot on the public website, an internal employee assistant for relationship managers, and instant AI case classification — cutting routing from hours to the moment a case is created.
TA Resume Screener — Semantic Hiring AI
An LLM screening platform that reads resumes the way a strict recruiter would — semantic evaluation against the job description, structured fit scores, and reasoning you can defend. Keyword-stuffing doesn't work on it.
03 — Operating System
From boardroom to production.
STEP / 01
Discover
Stakeholder workshops, process mapping, and discovery sessions that surface what the business actually needs — not what it first asks for.
STEP / 02
Design
Solution architecture, user stories, acceptance criteria, and scope-gap analysis — specs precise enough to build from, because I build from them.
STEP / 03
Build
RAG pipelines, agent orchestration, prompt design, full-stack apps. Prototype fast, evaluate hard, iterate on real model behaviour.
STEP / 04
Ship & Measure
Production rollout, adoption tracking, impact metrics. 25% faster resolutions and 35% efficiency gains aren't estimates — they're measurements.