Goli Bhargav

About

About

I build AI systems that actually run in production — not pilots, not slide decks. Multi-tenant. Governed. Tied to measurable business outcomes. Today that means leading two production AI platforms at All My Sons Moving & Storage: a call-analysis SaaS that scores live sales conversations against an agent-performance KPI framework, and an internal MCP-based platform that standardizes how employees and applications access LLM-powered tools across operations, safety, and HR. Both shipped, both in active use, both built on the patterns I’d reach for again — LangChain / LangGraph orchestration, RAG over Atlas Vector Search, MCP for tool surfaces, Bedrock for managed model access, multi-tenant isolation as a first-class architectural concern.

The architecture side runs in parallel. I lead the company’s enterprise data platform RFP across Snowflake, Databricks, and Microsoft Fabric — vendor evaluation, POC scoring, TCO modeling, and stakeholder alignment across data, finance, operations, and executive leadership.

What’s worked across these initiatives: holding the architect, builder, and governance lenses at the same time. Architecting a reference pattern in the morning, writing the Python eval script for it after lunch, and defending the design choices to a board the next week — usually all on the same platform. That combination is the thing that keeps these systems running in production rather than stalling at the pilot stage.