HL7v2, FHIR, X12 EDI, and APIs across provider and payer systems, from the first data contract through ongoing operations.
Independent projects for healthcare data exchange, file validation, and cluster operations. Try the browser tools or read the source; the full list is on the selected-work page.
Parses HL7v2, X12 EDI, and CSV into semantic events such as patient_admit and lab_result, then routes them through configurable workflows with FHIR R4 output. The playground runs the real kernel in your browser as WebAssembly.
Go · Healthcare · HL7v2 · FHIR
One static binary that gates outbound X12, HL7v2, EDIFACT, delimited, and fixed-width files in CI, with every X12 rule cross-referenced to the TA1 or 999 code a trading partner would return.
Go · Healthcare · X12 · HL7v2
Kubernetes-native GPU scheduling and model serving across vLLM, Ollama, and llama.cpp. It runs the inference behind this site.
Go · Kubernetes · LLM · gRPC
A Go daemon that manages MCP servers and agent memory, plus Mills, the gated pipeline that plans, implements, tests, and merges work on the projects above.
Go · MCP · Kubernetes · GitLab
The tools above are maintained with Mills, an agent delivery pipeline I built and operate: its gates refuse work that fails a check, and escalations are counted, not hidden.
flexinfer
merged · 2nd try · $3.42 at list price
fi-fhir
merged · first try · $5.51 at list price
Last 30 days, across every repository Mills works in: 400 of 832 runs merged, 93% of gate evaluations passed, and 408 escalated.
The factory that maintains these tools, the GPUs that serve this site, the cluster underneath, and the MCP servers my agents can reach. Each row reads the same sanitized public API as the page it links to and says whether its figure is live or a saved snapshot.
The Mills delivery pipeline: every run's stages, gates, retries, and cost over the last 30 days.
Open models on self-hosted GPUs behind an OpenAI-compatible gateway, with the engine and hardware of each one that is warm.
The K3s cluster that serves this site: node readiness, GPUs, and the health score FlexDeck computes from them.
The Go MCP servers Loom Core exposes to coding agents, mapped by category with the tools each one lets an agent call without asking.
I work across implementation, partner coordination, and ongoing support.
HL7v2, FHIR, X12 EDI, and API integrations across provider and payer systems.
Experience in interface support, API engineering, architecture, and team leadership
fi-fhir parses healthcare data into structured events; edilint checks EDI and flat files before delivery.
Independent Go projects with source, tests, and documentation
K3s on Harvester, Flux GitOps, Prometheus/Grafana/Loki, self-hosted GitLab CI, and GPU inference.
The homelab behind my independent projects and this site
Practical notes from healthcare integrations, platform reliability work, and cross-functional delivery.
I put fi-fhir and edilint through Mills for the first time. It merged work I did not write, then exposed a test gate passing on the wrong tree and fixed it against itself.
ai-agents · mills · loom · fi-fhir
A design pattern for supervised AI investigations: scoped read access, reviewed drafts, and failure tests before expanding permissions.
ai-agents · llm · guardrails · mcp
Four defect classes that healthcare interchange parsers can accept silently, with synthetic samples, real linter output, and the pre-send gate that catches them.
healthcare · edi · x12 · validation
A 90-day rollout for AI-assisted development: start with bounded habits, measure the work, and standardize only after the team has evidence.
agents · ai-assisted-dev · adoption · team-practice
The selected-work page links to source, the resume covers the history, and email is the fastest way to reach me.
