Full-Stack Development & AI-Native Engineering
Senior engineer and technical lead with a decade at Hughes Network Systems, building agentic AI systems daily with Claude Code, Cursor, and MCP — and shipping and teaching production software for decades.
For the last decade I've been a senior engineer and technical lead at Hughes Network Systems — architecting systems, reviewing code, and mentoring more than ten engineers, several of whom were promoted into senior roles. I was the first there to adopt agentic AI programming, and I built and ran the internal education that brought it to the rest of the team. That sits on top of decades writing production software — Ruby on Rails, Angular, React, TypeScript, Node.js, and AWS — mostly at scale inside established engineering teams.
I've been early to this before. In 1996, I built a self-learning weighted-logic system on personal hardware — years before "machine learning" was a term anyone used, while taking whatever engineering work I could find as a new immigrant. I was working with DALL·E in 2022, then ChatGPT from launch, and by April 2023 I was paying for production OpenAI API access and building with it at work — before most engineering organizations had even decided whether employees could use these tools. That same pattern carried into agentic programming: I was first at Hughes to adopt it, and I built the internal training that brought the rest of the team with me.
Alongside that, I build agentic AI systems by hand: a mesh-network security architecture for distributed embedded hardware, a SaaS app shipped end-to-end with Claude Code and Cursor, and generative-AI pipelines trained on custom models. Explaining new tools clearly, so other people can build with them, is something I've done publicly since the first wave of AI image models.
Independent engineering work, built and shipped outside of day-job employment.
38 repositories, 174+ stars on GitHub →
Alongside my engineering roles, I've built a technical-education practice at Geekatplay Studio — 3,000+ tutorials and a 96,000+ audience across English and Russian YouTube channels — with training used by artists at studios including ILM, Blizzard, DreamWorks, and Ubisoft.
I was teaching prompting and workflow for the first wave of AI image models back in 2022, and I'm publishing on agentic and ComfyUI workflows now — same practice, same audience, three years apart. Explaining new tools clearly, so other people can build with them, is something I've done publicly for a long time.