Tata Consultancy Services
Developer, AI & Data
2024 - Present
- Own a core area of the agentic component at TCS WisdomNext AI, delivering capabilities across 5 areas: agent orchestration, tool execution, RAG, automation, and developer tooling.
- Engineered agent SDK and reusable tool framework using LangChain/LangGraph, supporting 20+ agent types and 15+ tools for 10+ developers.
- Established a hybrid RAG stack with OpenAI embeddings, FAISS, and BM25, using 3-query expansion, 60/40 semantic-lexical fusion, 12-candidate reranking, and 2-pass synthesis with context expansion.
- Built natural-language-driven workflows for automated agent and tool provisioning, reducing integration effort by an estimated 1–2 hours per integration.
- Resolved production issues across 7 areas including LLM failures, agent loops, API and tool errors, concurrency, latency, memory, and data consistency, improving platform reliability.
- Optimized agent initialization and document-ingestion bottlenecks, delivering an estimated 10-15% improvement in execution efficiency.
- Mentored 10+ developers through architecture reviews, debugging, and code reviews, enhancing team capabilities.
