At AssetSense, I built Zen, an LLM-powered plant assistant using Python, GPT-4o-mini, Neo4j, and orchestration. I implemented rate limiting and concurrent request handling, reducing p95 response latency by about 40%.
I also designed and implemented I2, a configuration-driven integration engine that runs XML workflows at runtime. It replaced system-specific adapters across 40+ plant sites and supports integrations through REST, SOAP, and stored procedures.
For Digital Plant, I designed a real-time Neo4j integration and synchronization pipeline for multi-tenant data, reducing query latency by about 80% versus Oracle-based retrieval. I also researched and architected a configurable plant-operations workspace for digital twin, fault detection, optimization, work orders, asset tracking, analytics, and alerting.
At AssetSense, I co-led a five-engineer team building Catalyst, designing its MVC architecture and contributing to full-stack development with Next.js, tRPC, Prisma, and Jotai. Earlier, as a Full Stack Developer Intern at Metaverse Ventures Private Limited, I maintained a collaborative document builder and implemented GeminiAPI content suggestions.
