/Essays · publications · latest first
Essays
22 pieces, 4 pillars, 3 series. Every claim carries a number.
Series SpotOn 5 published
01 What I'm Building, and Why Alone 02 The Co-Founder That Builds Too Fast 03 A Roadmap One Person Can Finish 04 The Two Seats a Harness Can't Fill 05 How I Run an AI Team of One Series AI-Written Firmware 4 published · 1 upcoming
01 How Far Can You Trust It? 02 Where It Breaks 03 233 Tests, Sonnet 68% vs Haiku 57% 04 How to Stop AI from Breaking Firmware 05 Introducing the Verification Gate
Series AI Agents in Embedded 3 published
01 Spec-Driven TDD 02 Context Engineering 03 The Full Pipeline 09 standalone essays
- 01 Embedded Dev LLM PCB Routing: Turning a Human's Fixes Into Rules I handed a 96% auto-routed PCB to a contract engineer. He did not just connect the last 4%, and that difference became 16 rules my agent can reuse.
- 02 Embedded Dev Per-Project AI Agent Harness: Side vs Production I run side and production projects at once, so the workflow has to differ. Most harnesses are fixed, so I built one that generates a harness per project.
- 03 Thoughts How Long Does Harness Engineering Have? Everything useful in a harness eventually gets absorbed into vanilla. What survives is only what the vendor structurally will not or cannot build.
- 04 Embedded Dev What loop do I need to keep improving my software with LLMs? Working hypothesis on whether I can wire my own project into an LLM-driven self-improvement loop.
- 05 Cyber Security The CRA Begins: Embedded Software's Center of Gravity Shifts September 11, 2026: the EU Cyber Resilience Act's first obligation hits. Not compliance work. A center-of-gravity shift in embedded software engineering.
- 06 Thoughts AI Finally Needs Hardware Engineers Inference overtook training in 2026. The edge is growing fastest, and the work that remains is the work AI cannot do.
- 07 Edge AI Offline Datasheet RAG: What Survived the Pipeline Default RAG breaks on hardware docs. Here is the pipeline that works offline: SVD parsing, structural chunks, hybrid search, multi-query decomposition.
- 08 Edge AI Hardware Docs for LLMs: grep vs RAG RAG destroys the cross-references that make hardware docs useful. Skip the pipeline: give the LLM your SVD and PDF as file tools.
- 09 Thoughts Why I Started EdgeLog What happens when you push LLMs, zero-trust security, and modern frameworks onto real embedded hardware? A research log from the edge.