At National Instruments - NI (now part of Emerson), I architect scalable multi-agent support-ticket workflows with LangChain, LangGraph, Azure AI Search, and vector databases. I also spearhead vLLM and SGLang inference microservices, optimizing GPU workloads with continuous batching and KV-cache reuse.
At Webtrigon Technologies Pvt. Ltd. & Nagu Technologies Pvt. Ltd., I directed engineering units and advanced MLOps platforms and GPU scheduling with Kubernetes, Ray, and Slurm. That work supported AI products that unlocked $2.5M in new annual revenue.
At Infilect, I engineered real-time ML backend systems and inference APIs sustaining 500+ TPS at 99% uptime. I optimized PostgreSQL and ArangoDB-backed APIs, achieving 80% faster queries and reducing server load by 40%.
Across my projects, I’ve architected distributed LLM inference systems and agentic document processing, including a system that processed 1,000+ documents and saved an estimated 1,000+ man-hours annually. My work spans production AI infrastructure, backend platforms, and engineering team development.
