TCS UK
- Built enterprise-scale cloud applications and AI-ready architectures
- Designed API-driven microservices and cloud-native data pipelines
- Led architecture discussions around AI integration and observability
Senior Software Engineer and Full-Stack Technical Lead with 15+ years building enterprise-scale cloud applications. Now architecting LLMs, RAG, AI Agents, LangChain, LangGraph and production AI systems.

Devesh Kumar Srivastava is a senior technology professional with extensive experience in software engineering, cloud architecture, enterprise application development, and AI systems engineering. He bridges 15+ years of full-stack delivery with deep hands-on practice in modern GenAI tooling.
He partners with engineering leaders to architect production AI platforms — from retrieval pipelines and vector stores to multi-agent orchestration and cloud-native deployment.
From research-grade agent design to enterprise CI/CD — built and shipped at scale.
Design enterprise AI systems, AI platforms, and scalable GenAI solutions.
Build production-grade Retrieval-Augmented Generation systems with evaluation.
Design LangGraph and Agentic AI workflows with orchestration & tools.
Help organizations adopt AI securely, responsibly, and effectively.
Build end-to-end AI-powered applications from interface to model.
Design scalable Azure and AWS AI platforms with CI/CD & observability.
Production-grade enterprise RAG platform with multi-agent search and citation grounding.
Reduced knowledge retrieval time and dramatically improved knowledge-worker productivity.
AI-powered document intelligence, search and Q&A on regulated enterprise data.
Document intelligence, search and conversational Q&A at enterprise scale.
Autonomous agents capable of retrieval, reasoning, orchestration and tool execution.
Self-directing workflows that compose tools and APIs at runtime.
Large-scale cloud platform for processing and distributing enterprise data flows.
Unified pipelines across business domains.
Agentic workflow system combining LangGraph, RAG and vector retrieval.
End-to-end automation of multi-step knowledge work.
Ingestion → Embeddings → Vector store → Retriever → LLM → Grounded answer
Planner · Researcher · Executor agents orchestrated via LangGraph state machine
Identity, data plane, model gateway, evaluation, and governance layers
Prompt design → Eval → Deploy → Monitor → Iterate with LangSmith
Tracing, token cost, hallucination detection, and feedback loops
"Devesh brings rare combination of enterprise architecture depth and modern AI craft. He shipped our RAG platform end-to-end."
"His grasp of LangGraph and multi-agent design unblocked a six-month roadmap in two sprints."
"A trusted senior engineer who speaks both boardroom and codebase fluently."
Available for AI architecture engagements, technical leadership, and enterprise consulting.