Udit Brahmadevara

Forward Deployed Engineer @ AI20 Labs

About

Most AI projects don't fail because the model is bad. They fail somewhere between the demo that impressed everyone and the version real people depend on. That gap is where I like to work. As a Forward Deployed Engineer, that mostly means sitting close to the problem, with the team, with the customer, with the messy data and building the pipelines and systems that carry an idea into production. In healthtech, I was the founding engineer on a HIPAA-compliant platform built from nothing. I worked on pipelines that pull meaningful signals out of unstructured medical reports, and just as much on the unglamorous parts: the cloud architecture, the encryption, the deployment workflows that keep patient data safe. When the data is that sensitive, "it works" isn't the finish line. These days I work in agentic AI and media, building the infrastructure that lets conversational agents do two things well, reach the right data, and pay for themselves. That means production MCP servers running at the edge, the rails that let agents transact, and the Go services, caching layers, and embedded React runtimes that keep all of it fast inside strict, unforgiving web environments. I've also learned the engineering is only half the job. I'm comfortable in the room where roadmap and pricing get decided, and I spend a lot of time translating between what we can build and what a customer actually needs, shaping pricing, smoothing developer onboarding, and stepping into technical sales conversations when it helps move things forward.

Country

United States

City

Buffalo

Industry

Computer Software

Skill

Go (Programming Language), Go-to-Market Strategy, Retrieval-Augmented Generation (RAG), Large Language Models (LLM), Continuous Integration and Continuous Delivery (CI/CD), Model Context Protocol (MCP), Google Cloud Platform (GCP), Cloudflare, Redis, OpenAI Gym, Matplotlib, NumPy, PyTorch, LLaMA, FAISS, LangChain, n8n, Pipelines, SQL, Quality Assurance

Experience

Forward Deployed Engineer

AI20 Labs

2025-8 - Present · 1 yr 2 mos

United States

Build and scale production AI systems across client platforms, and contribute to Go-To-Market strategy, enterprise pricing, developer onboarding, and client-facing technical sales. Took a HIPAA-compliant healthcare platform from first commit to launch. Owned cloud architecture, automated CI/CD pipelines, and end-to-end PHI security, alongside the AI pipelines that parse unstructured medical reports into structured, longitudinal biomarker histories. Build the infrastructure that lets conversational AI agents integrate with proprietary data and generate revenue. Production Model Context Protocol (MCP) servers on Cloudflare Workers (TypeScript, Hono, Durable Objects) that expose client APIs as purpose-built agent tools without exposing credentials, plus a backend-driven server registry so new deployments ship without a frontend release. In-chat agentic ad delivery and HMAC-secured payment rails for agent-to-agent commerce: signed request verification, pre-flight allowance checks, 402 payment-required responses, and real-time usage metering. Go microservices on MongoDB and multi-tier Redis caching, including the cache-invalidation layer that keeps client configuration consistent across services. Retrieval and ingestion pipelines, recurring content ingestion, cross-agent syndication, and context-window tuning for RAG agents. Embedded React widget runtimes that render inside third-party sites under strict Content Security Policies.

Software Engineer Intern

Aude.ai

2024-4 - 2024-7 · 4 mos

Software Engineer

ShanMukha Innovations Private Limited

2024-2 - 2024-3 · 2 mos

AI Developer Intern

Sony India Software Centre

2023-6 - 2023-9 · 4 mos

Bengaluru

Worked in Sony India Software Center for 3 months in the AI/ML domain. Worked on understanding client requirements and doing research and POC development for multiple projects. Worked in computer vision and image processing.