Manan Jain

About Manan Jain

I build AI systems that actually work in production — not just in notebooks.Over the past ~2 years at TCS, I\'ve designed and shipped GenAI/NLP systems end-to-end: from data pipelines and retrieval architecture to model integration, evaluation, and cloud deployment. My focus is on systems that are measurably better — higher answer precision, lower hallucination rates, faster latency, and trustworthy outputs at enterprise scale. Adobe Document Cloud Hackathon Winner National AI Hackathon Winner — Govt. of India Google Cloud Gen AI Finalist — Top 10 of global participants🥈 ICEA AI Challenge Finalist What I specialize in → RAG Architectures — hybrid retrieval (BM25 + dense vectors + rerankers), adaptive chunking, grounded citations, cross-encoder reranking, MMR diversity→ Agentic AI — planner/critic/executor pipelines, tool/function calling, policy guardrails, failure observability→ LLM Fine-tuning — LoRA/few-shot workflows, eval harnesses, A/B testing, prompt iteration→ Document AI — OCR → NLP pipelines, structured JSON extraction, human-in-the-loop review→ Production Delivery — REST APIs, autoscaling, p95 latency optimization (caching, batch inference, embedding precomputation), CI/CD Delivered outcomes ↑ Boosted answer precision via hybrid retrieval + query understanding + cross-encoder reranking; reduced hallucinations with citation-first prompts and output verifiers↓ Cut review time ~30% through agentic planner/critic/executor workflows and guardrails↑ Improved p95 latency through caching, batch inference, and embedding precomputation↓ Reduced data access time ~50% via custom MCP server exposing internal microservices to LLM agents↓ Cut manual debugging ~70% with Agentic RCA system for failing APIs Stack GenAI: LangChain · LlamaIndex · LangGraph · OpenAI · Gemini · Ollama · Hugging FaceRetrieval: FAISS · ChromaDB · Pinecone · BM25 · cross-encodersML: PyTorch · Scikit-learn · LoRA fine-tuning · embeddings · NER (GLiNER)Cloud: AWS (Lambda, ECS/EKS, S3, CloudWatch) · GCP (Vertex AI, Cloud Run)DevOps: Docker · GitHub Actions · Grafana · Prometheus · CI/CDBackend/Full-Stack: Python (FastAPI · Flask · Django) · MERN · REST APIs · Neo4j Certifications Google Cloud Certified Professional Cloud Architect AWS Certified AI Practitioner Google Generative AI Leader Cisco CCNA Cyber Ops · ServiceNow CAD & CSAOpen to roles where applied GenAI and retrieval deliver measurable business value.Portfolio & GitHub available on request — let\'s connect.

Experience

1. Ai Software Engineer

   Tata Consultancy Services

   Nov 2024 — Present · Mumbai, IN

   Developing full-stack applications using the MERN stack for dynamic web solutions.Integrating AI features to enhance functionality with machine learning and NLP techniques.Collaborating with teams to build scalable and optimized solutions.

Education

• JECRC University

  Bachelor of Technology - BTech, Artificial Intelligence and machine learning

  2020 — 2024

• JECRC University

  Bachelor of Technology - BTech, Computer Science