Infrastructure and platform engineering.
I'm Jeff Doto. I build the infrastructure AI agents run on: runtimes, MCP tool layers, model gateways, and the observability to see what an agent actually did. 18+ years in software, 7+ running platforms at scale: GPU inference, Kubernetes-based platform modernizations, cross-cloud migrations. Open to full-time roles building agent platforms.
Agentic infrastructure
Agents as first-class workloads: MCP tool servers on Kubernetes, orchestration with LangGraph and kagent, model routing and cost controls through an LLM gateway, and the observability to know what an agent actually did. Also the reverse: agents that run on-call triage and GitOps changes with a human in the loop.
Observability
Make your systems legible. OpenTelemetry strategy and rollout, metrics, logs, and tracing that engineers actually use, agentic triage and synthetic monitoring, and alert quality.
Kubernetes
Cluster architecture, cloud-to-cloud and legacy-to-Kubernetes migrations, GitOps delivery, and GPU workload orchestration.
Cost optimization
Cut spend without cutting capability: AWS bills, Kubernetes right-sizing and Karpenter-driven autoscaling, GPU utilization, and observability spend.
Platform engineering
Paved paths for engineering teams: CI automation, custom tooling, feature flagging, and performance testing frameworks.
Infrastructure at Stability AI. After hours: a Proxmox homelab that hosts most of my experiments. Lately: LLM and agent tooling for infrastructure work, RAG pipelines, and audio DSP side projects.
Hiring for agent or platform infrastructure? Email me. I also take on a small number of consulting engagements.