Make the boundary visible
Good systems name ownership, failure modes, permissions, and the places where people need to make a decision.
About
Engineering leadership, hands-on systems work, and a few projects that got out of hand.
I'm an engineering leader and builder based in Denver, Colorado. My career has moved through military communications, aerospace, startups, AI platforms, and security.
Most of my work sits at the point where a difficult technical idea has to become a real product. That usually means building the team, defining the architecture, creating a way to measure quality, and figuring out how to operate the system once customers depend on it.
Working style
The same habits show up in leadership, platform work, and the writing here.
Measure before arguing
Evals, traces, regression baselines, and product feedback beat long debates about what probably works.
Keep the operator in the loop
AI and platform work fails when the demo hides what production users will have to debug later.
Experience
Keeper Security
Director of Engineering, AI & Threat Analytics
I built and lead a 15-person AI and threat analytics engineering team. The work spans AI platforms, LLM and API gateways, MCP, governance, threat analytics, privacy-preserving inference, quality, planning, and regression testing across products used by 30M people.
Vody
Head of Engineering / founding team
I joined pre-seed and owned engineering end to end. We took the product from concept into production, worked directly with customers and investors, and built multimodal, embedding, and classification systems around real product constraints.
Lockheed Martin
AI/ML, edge, data, manufacturing, and platform engineering
My Lockheed work moved through AI/ML platforms, secure 5G and edge systems, distributed model serving, data infrastructure, CI/CD at large scale, engineering mentorship, and manufacturing systems.
U.S. Army / White House Communications Agency
Critical communications and high-stakes operations
That part of my career is one reason I care about reliability. Some systems have to work under pressure, with clear ownership and fewer excuses.
Professional work and personal exploration
The two sides stay separate, but the pattern is the same: make the system legible, keep the operator close, and leave room for the thing to become useful.
professional focus
AI platforms, API gateways, model routing, threat analytics, governance, access control, reliability, and engineering systems that teams can operate.
personal exploration
Agents, MCP tools, local model inference, browser AI, WebGPU, small software products, Kubernetes, home automation, and whatever the homelab needs next.
Technical stack
- Languages
- Rust, Python, TypeScript / JavaScript, Go, Java, SQL
- Platforms
- AWS, Azure, GCP, Docker, Kubernetes, Terraform
- AI and data
- LLM gateways, model serving, MCP, agent systems, evaluation, MLOps, PostgreSQL, Redis / Valkey, Kafka
- Engineering systems
- GitHub Actions, GitLab CI/CD, OpenTelemetry, Prometheus, REST, gRPC, GraphQL