Headline

Agentic AI Engineer | Multi-Agent LLM Systems | RAG · Weaviate · LangGraph | Shipping Production GenAI @ Capgemini

About

As a Software Engineer at Capgemini since September 2024, I specialize in developing production-grade multi-agent AI systems, focusing on retrieval-augmented generation (RAG), LangGraph orchestration, and LLM guardrails. My work has contributed to improving document retrieval accuracy and reducing LLM hallucination rates through advanced techniques like hybrid vector reranking and NLI-based fact-checking. A graduate of G.H. Raisoni College of Engineering (GHRCE), I bring expertise in LangGraph, RAG, and multi-agent systems to enterprise-scale Generative AI deployments. My professional interests include LLM safety, guardrails engineering, and creating AI systems that are both capable and reliable in production environments.

Location

Nagpur

AI stack

hugging face

langgraph

langsmith

llmapi

openai

githubcopilot

Work experience

Software Engineer at Capgemini, Sep 2024 - Present · 2 yrs 1 mo, Navi Mumbai, • Built production multi-agent RAG pipelines using LangGraph and LangChain with Google Gemini API, improving enterprise document retrieval accuracy by ~40% via semantic chunking, transformer embeddings, and hybrid vector reranking on Weaviate and ChromaDB. • Reduced LLM hallucination rates by ~30% across production agents using NLI-based fact checking (DeBERTa), chain-of-thought prompt engineering, and context-window management strategies. • Architected a LangGraph guardrails gateway in FastAPI intercepting all LLM input/output — blocking prompt injection, PII leakage, and adversarial attacks across 8 attack categories; integrated Microsoft Presidio + spaCy for real-time PII detection and redaction. • Deployed AI microservices on Kubernetes with HPA (auto-scales at CPU > 70%) and built a Jenkins CI/CD pipeline with a Ragas quality gate (faithfulness ≥ 0.75), reducing low-quality model deployments to near zero. • Built end-to-end LLM observability using structured logging, distributed tracing, and LLM performance metrics — reducing MTTR by ~35% across all deployed agents.

Game Developer at MindYourLogic, Jun 2023 - Dec 2023 · 7 mos, Nagpur, Maharashtra, India

Intern at InternPe, Apr 2023 - May 2023 · 2 mos

Python Developer at Pantechelearning, Mar 2023 - Apr 2023 · 2 mos, Remote

Education

G.H. Raisoni College of Engineering(GHRCE), Nagpur