# Neel Barmecha > Neel Barmecha is a forward deployed AI builder and New York State certified EMT based in New York City. He deploys AI agents into live operations, with a focus on healthcare, and is open to forward deployed engineering, AI implementation, solutions engineering and AI operations roles in New York or remote. He is available to start immediately. ## Now - Velanir. Selected from 350+ applicants for a 20-seat Forward Deployed Expert (FDEx) cohort focused on healthcare. His capstone is building AI agents for a real client, and the cohort ends in a contract Forward Deployed Expert role. - Planisphere. Hands-on contributor on a live clinic EHR migration from Tebra to Athenahealth. He designed the AI agents for the data migration, drives clinical and front-desk staff through go-live, and helped recover information for 30+ patients after a third-party deletion incident. - Mudita Studios. Workflow design and implementation. Runs discovery with operators, built Signal (a sourcing and enrichment service with Claude reasoning over three data providers, exposed as a REST API and an MCP server) and its evaluation layer, and helped manage the team delivering a five-workflow agent system. - Styloire. CTO, sole engineer and product owner. Built the platform from scratch, ran discovery with 25+ stylists, and replaced the 100 to 500 manual outreach emails per project with one personalized send. ## Earlier - Klauza (2021 to present). Founder of a contract enforcement and collections tool for freelancers. MVP shipped with tiered pricing at $80, $350 and $450 a month, later rebuilt on the Claude API to flag predatory clauses. - EverPage. CTO. Shipped a social reading app to the App Store as the sole engineer, with 500+ people on the launch waitlist. - The Common Collective (2024). Founder of a members community that reached 300 members in its first week. - Google Global Youth Advisor via Canvas8 (2026). One of roughly 100 advisors worldwide. ## Builds - [trustlayer and mcp-trustlayer](https://github.com/neelbarm): open-source Python library and MCP server that decide when a model's output can be acted on, with 19 tests. - [Plancheck](https://plancheck-neel.netlify.app): a document-intake agent that holds the fields it can't verify for review. - [Go-Live Academy](https://golive-academy.vercel.app): role-based go-live training for clinic staff switching practice software. - [Company builds](https://neel-company-builds.vercel.app): unofficial prototypes built for companies he wanted to work with, using synthetic data. ## The 0.94 problem A value can come back from an AI model at 0.94 confidence and still be wrong. His fix ties every field to its verbatim source, runs a second independent pass against that source, and routes anything unverified to a person. ## Skills Python, TypeScript, React, Node, PostgreSQL, Claude API, Model Context Protocol, agent design, tool calling, guardrails, evaluation of AI output, customer discovery, requirements, go-live and change management. Languages: English, Spanish, Hindi. ## Education and credentials - B.S. Biology, The City College of New York (CUNY), 2026. - CUNY School of Medicine BS/MD program (biomedical sciences), 2021 to 2024. - NYS EMT, 2,000+ hours in the ER. - Claude Certified Architect (Anthropic). Velanir FDEx certification, in progress. Ten Google AI and Cloud certificates. ## Contact - Email: neelbarmecha5@gmail.com - LinkedIn: https://www.linkedin.com/in/neel-barmecha - GitHub: https://github.com/neelbarm - Site: https://neelbarmecha.netlify.app/