Swarnit Wayal

Architected and deployed an end-to-end Agentic AI system using LangGraph and Google ADK to automate the analysis of complex US mutual fund documents. I built multi-agent orchestration pipelines that decomposed extraction tasks into structured steps, integrating retrieval-augmented generation (RAG) and vector search (FAISS/Pinecone) to ensure high precision. This system improved document processing throughput by 3x and reduced manual effort by 40%. I also engineered React-based human-in-the-loop interfaces to facilitate real-time validation and model feedback.

Designed and implemented high-performance batch and streaming data pipelines using Kafka and Python to support real-time inference and large-scale data processing. I optimized SQL data models for high-volume datasets, significantly reducing system latency and improving query performance. To ensure production reliability, I established automated CI/CD pipelines and monitoring frameworks using Grafana to track model accuracy and performance drift. The implementation of these scalable back-end services contributed to over $150,000 in annual operational cost savings.