Software engineer focused on AI systems and realtime infrastructure. At Amazon worked on the eval stack for Rufus (10B$ in incremental sales), shipped manufacturing AI agents & MCP servers at AWS and built ultra low-latency video AI infra at FOX. Also contributed to end to end low-latency 3D live streaming.

Building realtime AI guidance for industrial and frontline workflows, combining live video AI with enterprise context. Looking to connect with people in manufacturing, field service and quality ops and those working on edge/on-prem AI deployment, vision AI or multimodal agents.​​​​​​​​​​​​​​​​

Currently working on real time video AI apps tor physical work

I’m Yuvraj Bhatia, founder and CEO of Privia AI. I build software that helps teams run AI in production with confidence by turning raw model usage into traceable metrics, monitoring, and actionable insights. I’m hands-on across product, design, and engineering: SDKs, APIs, data pipelines, and dashboards. I’ve also worked as an Undergraduate Teaching Assistant and enjoy explaining complex systems in simple terms. Right now I’m focused on shipping v1, onboarding early users quickly, and iterating with real production feedback.

AI reliability and observability, LLM evaluation and monitoring, production tracing and telemetry (OpenTelemetry style), data pipelines and real time analytics, developer tooling and SDK design, infra choices (Postgres, ClickHouse, Kafka), B2B SaaS growth, pricing and onboarding, and connecting with engineers and founders building applied AI products.

Building Privia AI, a reliability and observability layer for LLM and ML apps. I’m shipping an end-to-end platform plus SDK for capturing production traces, aggregating metrics (latency, error rate, costs), and surfacing issues with dashboards and alerts. I also prototype integrations (Slack, Linear, Postgres or ClickHouse) to speed up customer onboarding.