Dmitrii S.

I'm a full-stack developer, 20 years of React and Node. Most of my work now is AI agents, LLM workflows and automation for Healthcare and MarTech companies, plus the systems that run them. The last AI pipeline I built serves 13,000 users. Where I've worked • Healthcare, four platforms between 2016 and 2025: dental imaging, workforce management for care providers, veterinary diagnostics (that one ran 32 months), and moving a US medical network off its legacy stack onto Azure. Clinical software is unforgiving about correctness and audit trails, so that is where I start. • MarTech, three clients since 2025 and two of them still running: an ORM agency's internal platform, attribution and real-time analytics microservices for a B2B ABM platform, and ad reporting at volume. AI I've shipped • A driving-theory exam app I built from nothing. React Native on iOS and Android, React on the web, Nest.js behind both, 13,000 users. Underneath it there's a RAG pipeline that writes and checks a library of 5,000+ questions against a syllabus the regulator rewrites every year, with AI image generation and Spanish/English translation feeding the same content flow. • An ORM agency that was running its whole operation out of Google Sheets. I moved it onto Supabase and PostgreSQL, built the app in Next.js and Node, then added AI flows on OpenAI and Edge Functions that took over monitoring and reporting people had been doing by hand. What usually goes wrong with AI projects The ones I've watched stall didn't stall on the model. It was auth, or getting at the data, or cost, or nobody being able to say what the thing did last Tuesday. That is ordinary engineering, and it happens to be what I've spent two decades on, a good part of it in regulated places where a security review is part of shipping. I'll also say when AI is the wrong tool for a job, since a plain script that costs nothing often does the same work. Stack: LangGraph, OpenAI API, Node.js, TypeScript, React, Next.js, Nest.js, React Native, PostgreSQL, Supabase, MSSQL, MongoDB, Azure, CI/CD. Send me the brief if you're automating a process or putting AI into a Healthcare or MarTech product, and I'll tell you how I'd approach it. I work hours that overlap the US and Europe, and I do the building myself.