AI Infrastructure & Cloud Solutions Architect.

I'm in the room when the architecture gets decided — and still there when it goes to production. Fourteen years designing enterprise systems — the last five focused on AI infrastructure: Kubernetes at hyperscale on bare-metal AWS Outposts, microsecond-latency cluster networking (Cilium / Multus / SR-IOV / Mellanox CX-7), and the kernel-level tuning — CPU pinning, NUMA awareness, IRQ affinity, isolcpus — that turns a general-purpose cluster into a deterministic real-time platform. I've taken that stack from AWS internal labs into globally-distributed production at four-nines availability.

Alongside the cluster work, I deploy Claude and other foundation models into regulated enterprise environments — designing RAG pipelines on AWS Bedrock, building evaluation harnesses that measure models against each other on real customer data, and guiding teams through the messy reality of getting AI past prototype.

Increasingly that starts earlier: technical discovery with senior stakeholders, scoping the proof of concept that de-risks a decision, and making the case in front of an exec audience. What makes that credible is the delivery record behind it — I'm not pitching an architecture I haven't had to operate. Hiring, mentoring and architecture reviews on the side.