Skip to main content
Tanmay Kayande

Tanmay Kayande

AI Engineer – Python, FastAPI, RAG & Agentic Systems

Talent
GurgaonMember since October 26, 2026
Agentic AILangGraphguardrailsLLM evaluationFastAPIPydantic v2Alembicmicroservicesevent-driven architectureSQLPostgreSQLpgvectorQdrantRedisSQLiteRabbitMQDockerAWSpytest
·

Hire this person

Send a job offer directly to this candidate

Agentic AILangGraphguardrailsLLM evaluationFastAPI

About

AI Engineer with 2.5 years building production Large Language Model (LLM) applications in Python/FastAPI, including a WhatsApp multi-agent platform for a leading life insurer. Experienced in RAG, LangGraph agentic workflows, tool calling, and LLM evaluation; focused on LLMOps reliability, observability, and cost.

Experience

System Engineer (Agentic AI & Backend Systems) - Tata Consultancy Services - Gurugram, India

(2024-05)

  • Built a production WhatsApp multi-agent AI platform (async FastAPI microservices) for a leading life insurer’s policy queries and service requests, handling 135,000+ API calls/day across 320+ concurrent sessions.
  • Cut per-token embedding costs by about 92% vs. OpenAI text-embedding-3-large by benchmarking OpenAI, Gemini, Qwen, and BAAI models and selecting Qwen3-Embedding-8B for the RAG pipeline (pgvector, Qdrant).
  • Achieved 93% retrieval accuracy on a 100-query eval set with hybrid keyword + semantic search and reranking.
  • Reached 96% routing accuracy with LangGraph multi-agent orchestration, Redis cache lookups, and intent classification; on low-confidence matches, users can raise a service request that is auto-routed to the right category.
  • Cut LLM API failures from 15–20% to near zero via retry/backoff middleware and prompt-injection guardrails.
  • Delivered WhatsApp notifications and status updates via event-driven architecture on RabbitMQ; stopped duplicate replies with correlation IDs and delivery tracking, and message floods with per-user queue limits.
  • Containerized services with Docker and set up AWS disaster recovery with a sub-5-minute recovery time.

Education

B.E. - Computer Engineering (Hons. in Data Science) - MES College of Engg., Pune (2019 - 2023)

Skills

AI EngineerPythonFastAPIRAG & Agentic Systems

Reviews

Related