Bitan Basu — AI Implementation Consultant · GenAI · Enterprise Technical Solutions & Consultation

AI Implementation Consultant.
Built on eight years of Enterprise Technical Solutions & Consultation.

I help enterprises identify where GenAI creates measurable business value, scope the implementation, coordinate with data scientists and engineers, and communicate ROI to leadership — without needing a translator on either side. Grounded in eight years as the technical bridge across HP Inc. (Team Captain, SEA/AU) and Replicon (Deltek) (Fortune 500 advisory, C-suite escalations) — translated into four live GenAI platforms shipped solo.

Currently shipping at getqualityos.com·Open to senior implementation roles

01 TriageOS 2026

Production-ready Solo build Agentic RAG

Agentic RAG ticket triage engine with self-correction.

Reads company SOPs and automatically classifies, routes, and drafts replies for incoming support tickets across 8 dimensions. When confidence is low, a second LLM (the Critic) diagnoses failure and triggers refined re-retrieval — achieving 80%+ first-attempt accuracy. 20-node n8n workflow, self-hosted vector pipeline, sub-2-second end-to-end latency.

n8n ChromaDB Groq · Llama 3.3 70B sentence-transformers Flask Supabase Slack Docker Oracle Cloud Caddy

Live demo ↗

Send a test ticket, get a classification in under 2 seconds — no login required

What's actually built

Agentic RAG with self-correction — Critic LLM diagnoses low-confidence outputs and triggers refined re-retrieval
8-dimension classification — category, urgency, sentiment, routing, confidence, reasoning, draft reply, SOP citations
Self-hosted vector pipeline — ChromaDB + Flask embedder (all-MiniLM-L6-v2, 384-dim) on Docker/Oracle Cloud. Data never leaves client infra
6-path convergence — all outcomes into a single audit log with conditional Slack alerting
Sub-2s end-to-end latency including Postgres write · structured JSON · per-client thresholds

02 QualityOS 2025–26

In production Solo build Revenue-ready B2B SaaS

Revenue-ready B2B SaaS for AI-powered call & ticket QA.

Live at getqualityos.com. Eliminates manual QA — audio in, evidence-backed scorecard out, in under 30 seconds. Competing with Klaus and MaestroQA at a fraction of the price. Subscription billing, persistent data, Zapier integration to 5,000+ platforms, full team management — all shipped.

Next.js 14 FastAPI Groq · Llama 3.3 70B Whisper large-v3 Hybrid RAG · BM25 Supabase Clerk Dodo Payments Vercel · Railway

Open product ↗

03 ClarityCS LangChain

Live demo Multi-agent system

Five-tool LangChain orchestration for CX workflow automation.

Pre-briefs senior agents before they touch a ticket. Scores resolution quality 1–10 against policy. Catches at-risk accounts before they escalate. Modular data layer designed for direct swap to live Zendesk / Freshdesk / JIRA APIs.

LangChain ChromaDB Groq · Llama 3.1 HuggingFace Embeddings Streamlit RAG

View code ↗

04 NexusCS 12 modules

Live demo Full platform

CS Operations visibility layer for managers and stakeholders.

Twelve modules from executive dashboard to ML pipeline viewer to RAG ops assistant. Built for board-room readability — not a chatbot, a platform. Maps directly to enterprise tooling (Freshdesk Automations, Salesforce Agentforce, Power BI).

Single-file HTML Vanilla JS RAG · ChromaDB-ready Executive dashboard 12 modules

Open platform ↗ View code ↗

AI Implementation Consultant.
Built on eight years of Enterprise Technical Solutions & Consultation.

Four live AI platforms. All shipped. All working today.

Agentic RAG ticket triage engine with self-correction.

Revenue-ready B2B SaaS for AI-powered call & ticket QA.

Five-tool LangChain orchestration for CX workflow automation.

CS Operations visibility layer for managers and stakeholders.

The bridge most organisations cannot fill from either side of the fence.

From the support floor to shipping AI products.

Hiring for the bridge between AI and enterprise clients? Let's talk.

Hiring for the bridge between AI and enterprise clients?