Senior Staff Architect · Platform & Agentic Engineering
Building the platforms engineers — and now agents — build on. Kubernetes architect, eBPF enthusiast, and agentic-systems builder with 15+ years transforming how teams ship software.
I'm a platform engineer who believes developers are our customers. The best infrastructure is invisible—it just works. Every platform decision should be made with empathy for the engineers who will use it daily.
Most recently I was Senior Staff Architect at GE HealthCare, where I designed a multi-region AWS platform and, in parallel, built agentic developer tooling: cross-assistant AI skills adopted by 40+ engineers, self-hosted LLM inference, and orchestrator/worker agent architectures. Before that I led Alaska Airlines' internal developer platform — Backstage with custom Go services over a multi-cloud Kubernetes fleet.
The AI work isn't a job requirement — it's what I do at home too. I run a personal lab where I build agentic systems: Agentisan, a dual-provider subagent architecture with MCP-served skills that I use daily across Claude Code and Codex; Harness, my own multi-agent coding pipeline; and tdaddy, a test-impact engine that tells agents exactly which tests their changes affect — and which code nothing exercises at all. I publish under github.com/nerds-run, and I'm the #4 contributor to floci, the 24k-star open-source LocalStack alternative.
I also run Nerds Run LLC, a technology consulting firm specializing in platform engineering, DevOps, Kubernetes, and eBPF expertise based in Wisconsin.
My journey started in HVAC (yes, really), which gave me a unique perspective on building systems that need to be reliable, efficient, and maintainable. U.S. Army veteran.
Architected the Digital Fulfillment Console, a multi-UI platform on AWS replacing Flexera — micro-frontends behind CloudFront, GraphQL BFFs on Lambda, Cedar/AWS Verified Permissions authorization, multi-region active-active (architecture accepted, implementation underway).
Built and launched a company-wide architecture documentation platform (Docusaurus), and staged DPS's disaster-recovery architecture from active-passive toward active-active.
Built cross-assistant agentic skills (Claude Code, Codex, Copilot) adopted by 40+ engineers, piloted self-hosted Qwen inference on vLLM, and cut golden-image CVEs 70–80% via distroless migration.
Led Alaska's internal developer platform initiative using Backstage with custom Go backend services. Architected multi-cloud Kubernetes infrastructure — roughly 80 clusters across Azure, GCP, and on-premises data centers — and established the ADR process the platform org now runs on.
Led infrastructure initiatives for the talent acquisition platform. Redesigned AWS account structures and a Transit Gateway networking backbone, and managed Kubernetes clusters and CI/CD pipelines for rapid deployment cycles.
Grew the Platform, DevOps, and SRE organization from a team of one to 10–12 engineers, and re-architected the AWS environment from scratch. Established cloud infrastructure standards, implemented GitOps practices, and drove platform modernization initiatives.
Architected cloud solutions for enterprise financial services. Focused on security, compliance, and scalability in regulated environments.
Built Unity's self-service logging and metrics platform and ran HA Kubernetes on AWS, migrating South America and Asia-Pacific clusters into a shared account.
Progressive roles from Software Engineer to DevOps Lead. Built cloud deployment platforms, containerization workflows, and partner integration systems.
Comprehensive developer portal using Backstage with custom Go microservices. Centralized onboarding, K8s deployments, service inventory, and SDLC across the organization.
Developed microservice operator and dynamic OIDC operator for automated Azure AD integration and workload identity management across multi-cloud infrastructure.
Led enterprise-wide Cilium CNI adoption alongside the deprecation of Istio service mesh. Developed automated tooling for zero-downtime cluster migrations.
Technology consulting firm specializing in platform engineering, DevOps, Kubernetes, and eBPF expertise. Based in Wisconsin.
Top contributor to the 24k-star open-source LocalStack alternative; upstream work powering enterprise local development.
Ansible collection taking an AMD Strix Halo machine from zero to an operational LLM inference server — vLLM/ROCm and llama.cpp/Vulkan backends, service modes, model management.
Docker image build-and-distribution SDK/CLI from the early container era; used in production at TUNE, published to PyPI.
Agentisan (dual-provider subagent architecture, in daily use), Harness (multi-agent coding pipeline), tdaddy (Go-native test-impact and dead-code analysis wired into agent workflows), all running against self-hosted models on my own hardware.
My mission is to build reliable and easy-to-use platforms driven by the requirements of the business and the development organization.
Freedom to innovate, freedom from impediments, and freedom to ship with confidence. The best infrastructure empowers developers to focus on what matters—building great products.
Interested in platform engineering, Kubernetes architecture, or consulting engagements?
abanna@nerds.run