Writing
Blog
AI agents, runtime safety, local LLMs, and what it looks like to run a one-person AI-operated holding company in public.
Browse the notebook
Start with a topic, or open the full archive.
real test outputMy 8B Model Failed a 400-Word Task
Three Llama 3.1 8B runs missed a 400-word floor. Here is the verifier-driven route that moved long-form synthesis to Gemma 4 26B.
Read the post- 5 min read
My 5090 benchmark was missing the field I needed most
A fresh Qwen3.5 9B run showed 84.94 tok/s, but the useful number was the 6,105 ms load phase. I added phase timings and capture time to the benchmark receipt.
- 6 min read
Search Old Results Before Publishing an LLM Test
An independent QA pass caught my second post about the same Ollama batch sweep. Here is the duplicate check I now run before publishing an LLM result.
- 6 min read
Build Local LLM Eval Data From Real Failures
I show how I turn failed local coding runs into replayable eval rows with the prompt, model output, tests, route, and verifier result intact.
- 6 min read
How I Keep LLM Results Valid After a Driver Update
A new GPU snapshot does not refresh an old benchmark. I show how I bind driver, runtime, workload, and timestamps to each local LLM result.
- 6 min read
Why I Benchmark Local LLM Input and Output Separately
My RTX 5090 runs show why model loading, prompt ingestion, token generation, and task checks need separate measurements before production use.
- 6 min read
How I Budget VRAM for Shared Local AI Workloads
My RTX 5090 measurements show how I budget VRAM for a local model, a real workload, and the GPU processes that must stay resident beside it.
- 5 min read
Why I Test 3 Workloads Before Sizing a Local LLM
One local LLM speed number hides the work behind it. My RTX 5090 sweep shows why short generation, long context, and code need separate rates.
- 6 min read
Local open-model agents just became a product category
LM Studio shipped Bionic, a full agent built on open models with local code projects, voice, and document work. The interesting part is not the app. It is what.
- 6 min read
Incident response needs a local model you already trust
Hugging Face ran its breach forensics on an open-weight model on its own hardware because hosted APIs refused the requests. Here is the lesson for builders.
- 6 min read
Your local LLM benchmark is probably lying to you
A local model pass rate can be true and useless at the same time. Here are the three ways local LLM benchmarks mislead you, drawn from real rows on my RTX 5090.
- 5 min read
Test Retrieval Before Your Local LLM Writes
I make my local LLM show its source plan before it writes. The preview gate catches weak retrieval while the fix is still cheap and easy to inspect.
- 5 min read
Why I Did Not Promote My Smaller Local Model
My smaller local model existed and ran, but it did not beat the baseline. Here is the promotion gate I use before changing a working local AI route.
The AI agent build notes
Real costs, real tools, no fluff. M-F when I ship, publish, or learn something worth sending.