Building, reading, and trying to become less wrong.
I try to learn what I can about applied AI, human memory, hiring, startups, and whatever feels unresolved enough to keep tugging at the thread. After years of successive venture-backed AI roles, I'm stepping away to study, build independently, and choose the next problem carefully.
This site uses distilbert-base-uncased in the background as a small nod to everyone who remembers when fine-tuning a small encoder felt like touching the frontier.
About
Hands-on leader across scalable AI systems and full-stack products.
I'm happiest close to the problem: understanding it properly, building the first version, and then making the loop better. My work spans model inference and scalable GPU orchestration, agentic systems backed by virtual file systems and collaborative editing, evaluation pipelines, and full-stack product surfaces.
I care about the team as part of the system: hiring carefully, protecting morale, developing engineers, and making it easier for people to exercise good judgement under real product pressure.
Get in touch if you're thinking seriously about AI, memory, learning, or the awkward work of deciding what to do next.
