Principles
The cross-cutting ideas each post and case study is tagged with.
6principles
Writing
I write about the systems I build and operate: integration contracts, incident recovery, platform decisions, and the evidence behind them.
Method
The method behind the writing
The principles and playbooks the posts and case studies are tagged with, and reading paths through them.
Playbooks
Processes a team can adopt, with inputs, an owner, steps, checks, and a handoff.
4playbooks
Reading paths
Ordered sequences through the archive, each aimed at one goal.
4paths
Browse by topic
Every post and case study with its principle and playbook tags, filterable by either.
20pieces
Delivery write-ups
Case studies
Delivery write-ups with metrics, stack choices, and the tradeoffs that mattered.
6 min read
Featured · AI Infrastructure
Operating a Kubernetes LLM Platform on Consumer GPUs
How I evolved a small AMD GPU cluster into a CRD-driven inference platform with an OpenAI-compatible boundary, evidence-based model promotion, and safe rollouts.
From the study, as written Sep 12, 2026
- Serving Contract
- OpenAI-compatible
- Production Lane
- 2 replicas
- Delivery Model
- CRD + GitOps
Status: In progress
Live system for Operating a Kubernetes LLM Platform on Consumer GPUs6 min read
Featured · Healthcare
fi-fhir: Profile-Driven Healthcare Integration
How fi-fhir separates feed-specific variance from parsing, semantic events, durable delivery, and workflow policy—and where the pre-1.0 platform still has work to do.
From the study, as written Sep 20, 2026
- Input Families
- 4
- Ingress Paths
- 3
- Release State
- Pre-1.0
Status: In progress
Live system for fi-fhir: Profile-Driven Healthcare Integration5 min read
Healthcare
Patient Matching When the API Contract Is Ambiguous
A de-identified field pattern for containing patient-match risk when filter and mode parameters interact in undocumented ways.
From the study, as written Sep 12, 2026
- Observed Conflict
- 2 parameters
- Auto-link Policy
- Fail closed
- Regression Proof
- Pinned fixtures
Status: Completed
6 min read
DevOps
Building an Operator-Focused Kubernetes Dashboard
How FlexDeck combines Kubernetes, Flux, CI, observability, and model state without hiding freshness, access boundaries, or the limits of a homelab control surface.
From the study, as written Sep 12, 2026
- Operator Surface
- K8s + GitOps
- Application Shape
- Go + SolidJS
- Public Boundary
- Read-only
Status: In progress
Live system for Building an Operator-Focused Kubernetes Dashboard
Field notes
Blog
Technical notes, incident lessons, and implementation patterns worth keeping.
revised Sep 12
Professional · 9 min read
The Factory Files a Ticket Against Itself: Two Repos, One Session, Three Gaps
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
revised Sep 12
Lab · 6 min read
The Factory Upgrades Its Own Foundation: An MCP Spec Marathon
MCP shipped a breaking specification revision. I wrote the gap analysis, filed it as factory backlog, and Mills started rebuilding the SDK it runs on.
mcp · ai-agents · loom · mills
revised Sep 13
Professional · 4 min read
The Agent Gets Read-Only Credentials: A Trust Ladder for AI in Regulated Systems
A design pattern for supervised AI investigations: scoped read access, reviewed drafts, and failure tests before expanding permissions.
ai-agents · llm · guardrails · mcp
revised Sep 12
Professional · 5 min read
The File That Failed With Zero Errors: A Field Guide to Invisible EDI Defects
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
revised Sep 12
Lab · 12 min read
Twin Life on a Doubled Pool: One Day of Canary-Driven Inference Engineering
A twin-lane canary promoted prefix caching, FP8 KV cache, and a larger context pool while catching five regressions before they reached the primary workload.
flexinfer · prefix-caching · fp8 · speculative-decoding
revised Sep 12
Lab · 13 min read
Five Voices, 200K Tokens, One Consumer GPU: Giving Simulated Minds a Whole Life to Remember
How five simulated minds share one 24 GB Radeon, keep 200K-token life histories, reuse them through prefix caching, and compact old memories without erasing continuity.
flexinfer · long-context · multi-agent · prefix-caching