Abhisek Bose

AI Product Architect · Senior PM · Kolkata, India

I design and build AI systems, not features. AI Product Architect with 8+ years delivering LLM-powered systems, agent workflows, and multi-tenant SaaS platforms across e-commerce, HR tech, and funnel ecosystems.

Emailiabhisekbosepm@gmail.com LinkedInView profile →GitHub@iabhisekbosepm

Pinned projects

custom_agent

Custom Agents is a terminal-based AI coding assistant that runs entirely on your machine. It gives you specialized AI agents — explorer, coder, and reviewer — that can read, write, search, and reason about your codebase. Think of it as your own local Claude Code, powered by any OpenAI-compatible API (OpenRouter, OpenAI, Ollama, LM Studio).

TypeScript★ 3

qa-auto-agentic

QA-Auto is an automated QA testing framework powered by LLMs and browser automation. It reads BDD-style feature files, executes all scenarios, and generates Cucumber-style HTML reports.

Python★ 2

claude-god-setup

Multi-agent orchestration system with 21 specialized agents, automated workflows, and cost-optimized model routing.

JavaScript★ 1

Deep-Learning-Interactive-Tutorial-For-Everyone

Deep Learning Interactive Tutorial - That's all you need to know

JavaScript★ 0

cortexbrain

CortexBrain is a B2B AI knowledge system that gives organizations a persistent, self-correcting "internal brain."

HTML★ 0

cckanabn

No description.

JavaScript★ 0

Skills

Prompt craft

Few-shot promptingChain-of-ThoughtSystem promptsPersona designOutput formatting

Context engineering

RAG architecturesVector DBsEmbedding strategiesToken budgetingLong-context summarizationMemory design

Review & hallucination

Hallucination detectionEval frameworks (Langfuse)LLM-as-judge rubricsConfidence scoringAdversarial testing

Tools, agents & security

Multi-agent orchestrationMCP serversTool/function callingClaude CodeCursorGuardrailsPrompt injection mitigation

Experience

AI Product Architect — Independent (2024 — present)

Built OutcomeOS — AI Readiness assessment platform with role-mapped scoring, a real-task simulator with hallucination plants, founder-led quarterly cohort, and W3C-Verifiable-Credential Skill Passport (Ed25519 signed).

Senior PM, AI Products — 8+ years across e-commerce, HR tech, and funnel ecosystems

Delivered LLM-powered systems and agent workflows in multi-tenant SaaS
Designed evaluation pipelines for production LLM features
Led 0→1 product launches and platform-level architectural decisions

Selected Projects

OutcomeOS (outcomeos.online) — full-stack AI readiness platform: Next.js 16, Clerk, MongoDB, Razorpay, Resend, Vercel cron, Ed25519 credentials.
Custom Agents — terminal-based AI coding assistant that runs entirely on local infrastructure.
QA-Auto Agentic — automated QA testing framework powered by LLMs.
Claude / Codex / Gemini God-Setup — multi-agent orchestration systems with 21+ specialized agents, automated review pipelines.

Method

I design and build AI systems, not features. Every shipped LLM feature has:

A scenario-based eval suite before launch
A hallucination-detection rubric (not vibes)
A defined token budget and context strategy
A fallback for when the model is wrong

Currently looking for

Founding AI product / engineering leadership roles at companies shipping LLM-native products to real customers. Bonus points if the team treats evals as a first-class engineering surface.

Reach out: iabhisekbosepm@gmail.com or LinkedIn.

Make your own — free →