K Srini

I’m an AI/ML Engineer with 3+ years at TCS building production GenAI systems, RAG pipelines, and multi-agent workflows. I work across the full LLM stack—from data ingestion and embedding pipelines to vector search, fine-tuning (LoRA/QLoRA), model serving (vLLM, AWS Bedrock), inference optimization, and evaluation frameworks.

What sets me apart is that I don’t stop at prototypes: I deliver production-grade systems with platform engineering discipline. I’ve designed FastAPI microservices, deployed on AWS EKS, and used Terraform IaC, CI/CD automation, and observability to own the full path from model development to reliable runtime behavior.

In my role, I’ve improved retrieval accuracy ~25% with hybrid search (BM25 + dense vector) and cross-encoder reranking, reduced RAG response latency 30–40% using Redis semantic caching and async execution, and set up rigorous evaluation pipelines (RAGAS and custom LLM-as-judge) with regression tests and MLflow-based experiment lifecycle management. I’m especially energized by building agentic workflows with LangGraph/CrewAI and making them measurable, testable, and cost-aware in production.