# Swarnava Dutta

## Identity
Name: Swarnava Dutta
Role: Senior Generative AI Engineer
Location: Kolkata, India
Website: https://swarnava.is-a.dev
Email: swarnavadutta.official@gmail.com
Book a quick call: https://cal.com/swarnava-dutta/quick-call
GitHub: https://github.com/swarnava-dutta
LinkedIn: https://linkedin.com/in/swarnava-dutta
Twitter/X: https://x.com/Swarnava_Duttaa

## Summary
Swarnava Dutta builds production-grade AI systems for search, copilots, workflow automation, and decision support. His work focuses on LangGraph workflows, RAG architecture, AI agents, Azure OpenAI, enterprise LLM systems, fine-tuning, and data-backed AI products.

## Core Expertise
- Multi-agent systems and LangGraph orchestration
- Retrieval-augmented generation and vector search
- Enterprise AI architecture and Azure OpenAI delivery
- LLM fine-tuning, LoRA, prompt engineering, and evaluation
- Data engineering foundations for AI workflows

## Services
- Multi-Agent Systems: Design agent workflows that coordinate tools, memory, approvals, planner/reviewer/executor roles, and guardrails. Best for teams replacing repetitive manual work with structured agent handoffs. Stack: LangGraph, MCP, tool calling, guardrails.
- RAG Architecture & Vector Search: Build retrieval systems that surface trusted context from documents, data systems, and business knowledge. Best for internal copilots, search-heavy products, and grounded Q&A. Stack: Azure AI Search, Pinecone, reranking, RAG evaluation.
- Model Fine-tuning & Optimization: Improve model quality, speed, domain fit, tone control, and consistency through dataset design, LoRA/PEFT, prompt shaping, benchmarks, and inference tuning.
- Enterprise AI Architecture: Turn AI pilots into secure, observable, maintainable platforms with clear data flow, deployment patterns, governance, monitoring, and team handoff.

## Projects
- GRIT: Adaptive sales enablement for regulated medical teams using RAG, MCP-style content updates, LangGraph, and skill-gap analysis.
- Talksmith: Voice-first data copilot that turns spoken business questions into Snowflake-backed SQL answers through a LangGraph workflow.
- IntervueRecall: SaaS workspace that transcribes interview recordings and extracts interviewer questions into a clean review list.
- InsightDesk AI: Enterprise knowledge assistant that routes questions across documents, tables, and hybrid evidence sources using Databricks and RAG.
- LinkedInfluencer: Local browser app that discovers live GenAI discussions, scores relevance, and creates LinkedIn-ready posts with LangGraph.
- FitScout: Recruiter screening assistant for resume ingestion, skill matching, evidence-backed summaries, and candidate shortlists.
- WhisperIt Legal Copilot: Privacy-conscious legal document assistant for contract review, clause summarization, and legal Q&A.
- OpsAnswer Hub: Shared HR and finance assistant for policy, payroll, and operational support questions.

## Collaboration Fit
Best-fit work includes enterprise AI copilots, RAG architecture reviews, LangGraph agent systems, AI workflow automation, LLM evaluation, and production AI implementation.

## Availability and Contact
Status: Available for full-time opportunities, contract projects, and freelance consulting.
Preferred contact: LinkedIn or email.
Quick call: https://cal.com/swarnava-dutta/quick-call
Response time: Within 24 hours.