Headline

AI Engineer | Agentic AI & Enterprise RAG | LangGraph · MCP · Azure OpenAI | Building production multi-agent systems |

About

I build production AI systems that reason over enterprise knowledge and take action: RAG pipelines, multi-agent workflows and LLM-powered automation. What I work on 🔹 Retrieval and grounding: Multimodal RAG on Azure AI Search and Azure OpenAI, combining dense-vector and BM25 retrieval with semantic reranking. This improved Precision@10 by 35% and cut query latency from 3.2s to 1.9s. 🔹 Agentic workflows: State-driven multi-agent systems in LangGraph, LangChain and MCP, with supervisor-specialist routing, ReAct reasoning, tool calling and human-in-the-loop controls. 94% tool-execution accuracy, and workflow completion time cut from 40 to 20 minutes. 🔹 Evaluation and safety: LLMOps with RAGAS and LangSmith evaluation gates, golden datasets and OpenTelemetry tracing, plus Responsible AI controls such as PII masking, prompt-injection defenses and RBAC-based isolation. 🔹 Applied AI in the enterprise: Copilot Studio agents automating IT service workflows, and NLP and ML models for insurance claims analysis, built with Finance and business stakeholders around measurable cost outcomes. I came into AI engineering from a data analytics background, so I design for measurable outcomes and not just demos. Stack: Python · LangGraph · LangChain · MCP · Azure AI Foundry · Azure OpenAI · Azure AI Search · FastAPI · Docker · AKS Open to: AI Engineer, GenAI Engineer and Agentic AI roles.

Location

Bengaluru

AI stack

antigravity

hugging face

langchain

langgraph

Tavily

Work experience

Associate at Cognizant, Sep 2022 - Present · 4 yrs 2 mos

Machine Learning Engineer at Cognizant, Sep 2022 - Present · 4 yrs 2 mos, AI Engineer, Cognizant Technology Solutions Sep 2022 to Present · Bengaluru, India Built production RAG systems, multi-agent workflows and LLM-powered automation for enterprise clients on Azure. Built state-driven multi-agent workflows in LangGraph, LangChain and MCP, with supervisor-specialist routing, ReAct reasoning, tool calling and human-in-the-loop controls. Reached 94% autonomous tool-execution accuracy, cut workflow completion time from 40 to 20 minutes and cut report turnaround by 25%. Architected multimodal enterprise RAG pipelines for 100+ page PDFs, text and images on Azure AI Search, Azure AI Foundry and Azure OpenAI. Hybrid dense plus BM25 retrieval with semantic reranking improved Precision@10 by 35% and reduced hallucinations by 40%. Engineered the retrieval, indexing and grounding layers of the NJX AI platform, serving 10,000+ monthly queries at p95 latency under 1.2s. Optimized production RAG through reranking and prompt-chain redesign, cutting latency from 3.2s to 1.9s and LLM inference cost by 35%. Operationalized LLMOps with RAGAS and LangSmith evaluation gates (Recall@K ≥ 0.92, Precision@K ≥ 0.88, Faithfulness ≥ 0.90), golden datasets, OpenTelemetry tracing and Azure Application Insights. Built Copilot Studio agents integrated with Power Automate, ServiceNow, Teams and Outlook, automating resolution for 500+ monthly IT tickets and cutting mean time to resolution from 4 hours to 1 hour. Built spaCy and Hugging Face NER pipelines that improved claim-processing accuracy by 25%. Developed XGBoost models that improved first-time-right claim settlements by 10%, with projected reprocessing-cost avoidance of $10.6M over five years based on Finance-agreed assumptions. Delivered FastAPI microservices with Docker, GitHub Actions CI/CD and AKS in a 30-member Cloud & AI team, and mentored junior engineers.

Education

Bachelor of Technology - BTech, Electrical, Electronics and Communications Engineering, Madanapalle Institute of Technology & Science

Bachelor of Engineering, Electrical, Electronics and Communications Engineering, Madanapalle Institute of Technology & Science