Open to opportunities

Vinod Polinati

@vinodpolinati

Message

AI Engineer shipping production agentic RAG systems with cost control and zero unvalidated outputs.

India

Message

What I'm looking for

I’m looking for remote (or relocation) work where I can build production-grade agentic RAG and reliable LLM systems—cost-optimized, safety-guarded, and shipped end-to-end (backend + mobile) with strong engineering ownership.

I’m an AI Engineer focused on shipping production AI systems end-to-end—from the model layer to backend services and mobile deployment. At a US sports analytics startup, I architected a cost-optimised Agentic RAG system that deflects 65–75% of queries from the LLM (about ~70% API cost reduction) while enforcing “zero unvalidated stat responses” in production.

Earlier, I built a fully air-gapped offline RAG system at ISRO using Mistral 7B, LangChain, and FAISS, improving retrieval speed by 40% and reducing response latency by 30% with semantic reranking and query-time index optimizations. I also containerized the solution with Docker to make deployments reproducible across classified research environments.

Beyond RAG, I’ve done applied ML work—like designing a CNN + ILP optimization model for warehouse box selection (92% accuracy, 25% packaging cost reduction)—and I enjoy hardening AI features for real users. My projects include production-ready multi-agent research with LangGraph and robust LLM interview automation with FastAPI, React, and CI/CD.

Experience

Work history, roles, and key accomplishments

Current

AI Engineer

Current

Statslane, Inc.

Dec 2025 - Present (4 months)

Architected a production Agentic RAG system (LangGraph, FastAPI, MongoDB) that deflects 65–75% of queries from the LLM, reducing Groq API costs by ~70%. Implemented hallucination guards for zero unvalidated stat responses and delivered an end-to-end Flutter mobile app as the sole engineer.

Agentic RAG LangGraph fastAPI MongoDB 3 Tier Query Routing Stats Aggregation & Caching Flutter

Applied AI Research Intern

Indian Space Research Organisation (ISRO)

Jan 2025 - May 2025 (4 months)

Built a fully air-gapped RAG system (Mistral 7B, LangChain, FAISS) enabling offline document QA over ISRO’s classified research corpus. Improved retrieval speed by 40% and reduced response latency by 30% using semantic reranking and query-time index optimizations, and packaged it with Docker for reproducible deployment.

Air Gapped RAG Mistral Lang Chain Faiss Offline Document QA Docker

Machine Learning Research Intern

AegionDynamic

May 2024 - Sep 2024 (4 months)

Designed a CNN with ILP optimization for warehouse box selection, achieving 92% accuracy and reducing packaging costs by 25%. Validated the model in a live warehouse environment and authored documentation to support future deployment and scaling.

CNN Warehouse Box Selection Computer Vision Technical Documentation