Resolving
the signal
I keep AI in production — reliable, secure, and cheap enough to run.
GenOps — production discipline for generative AI. Ops rigor, aimed at LLMs and agents.
I thrive on the edge of innovation.
Systematically exploring, testing and hardening emerging technologies — then making them survive contact with production, at scale, under regulation.
Anyone can prompt a model. Very few can run it for 60 million users without it falling over — or bankrupting you. That gap is my job.
GenOps // production discipline for generative AI — Ops rigor applied to LLM & agent systems.
SERVE
LLM & vision-model serving — throughput and latency at 1.2M inferences/day. I make generative AI cheap enough to keep running.
ORCHESTRATE
Agent frameworks and data pipelines — the substrate under every RAG, eval and training loop. Current role: DataOps @ SolarEdge.
HARDEN
Nine years keeping money-moving fintech at 100% uptime — PaySafe, Nexo, Trading 212, Delasport. PCI-DSS. Now aimed at LLMs and agents.
The stack I run.
Money-moving, zero-downtime, regulated.
Notes from the frontier.
An outbound feed on how production AI actually behaves. Publishing soon.