About
Alireza Rezvani.
CTO and AI engineer in Berlin. 25 years turning hard problems into shipped software — and, lately, into tools and writing that help other people do the same.
I started as the first engineer on small teams. I've shipped at 2am, and I've sat across from a board explaining why we didn't. Both taught me more than any single win.
Twenty-five years later I'm CTO at a Berlin AI MedTech company, building systems that help people move safely and live better. Along the way — healthtech, AI, developer tooling — I've shipped products people use every day and quietly buried the ones that didn't. The failures taught me the most: how to read a hard decision early, and how to stand back up when one goes wrong.
These days my work points at agentic AI — building agents that actually ship, and the tooling around them. My open source has grown past 22K+ GitHub stars (345+ agent skills and 80+ plugins across 13 coding tools), and I research the harder questions underneath: computer vision, reinforcement learning, model fine-tuning, and corporate LLMs.
Most of what I learn, I write down. My writing reaches 7,000+ engineers on Medium and Substack, and I founded nGinity, a publication for people who ship with agentic AI rather than collect tool demos. The thread through all of it is teaching the next generation of CTOs and founders how to make good calls — and how to fail well. My aim is simple: extend what people can do, and help these systems learn from good human judgment.
The short version
CTO · Berlin AI MedTech
Leading engineering for AI that analyzes human motion and keeps people safe. Agentic systems, real stakes, real users.
Writing & open source
345+ open-source agent skills (22K+ GitHub stars) and writing on AI engineering read by 7,000+ developers.
Where AI learns from us
Computer vision, reinforcement learning, model fine-tuning, and corporate LLMs — making the most of human-level expertise, not replacing it.
Building & leading
From first engineer to technical leader across healthtech, AI, and developer tooling — at startups and at scale.
ISO 27001 · ITIL4
Certified in information security and service management — I care that AI systems are safe and auditable, not just functional.
Receipts over rhetoric.
Specific numbers, shipped code, real users. No "industry-leading", no demos that collect dust.
Teach, don't gatekeep.
The point of 25 years of scar tissue is to hand it forward — so the next CTO or founder skips a few of my mistakes.
Compliant by default.
AI that's safe and auditable from day one — because that's the part most people skip until it bites.