I build AI systems,
then write down what the numbers say.
I'm a Senior AI Engineer. Outside work I run small, pre-registered experiments on open AI models on one 16 GB graphics card: fine-tuning, calibration, quantisation and serving. I publish the numbers, including the results that didn't hold up.
Measured, not asserted.
The Same Coding Model Improved When Its Harness Changed
In the separate n09-n16 cohort, Patchsmith v5b/v6/Aider passed 13/19/23 of 24 for gpt-oss, 14/15/17 for Qwen3-Coder, 24/24/23 for Qwen3.6 and 18/17/23 for Edge0. Edge0 v5b is not blind; cohorts are never pooled. Two simultaneous v6 changes and unequal budgets prevent a causal harness ranking.
Expert Caching Made Tokens Faster. Coding Throughput Did Not Improve Clearly
For Qwen3-30B-A3B Q4_K_M on RTX 5060 Ti 16 GB, a 6,000 MiB cache raised later-request decode speed from 30–33 to 41–50 tokens/s, while verified coding throughput was 12.35 versus 12.93 passes per task-hour; correctness equivalence is not established.
A 120B Model Fit on My 16 GB Card. Reliable Coding Was Still Unresolved
Edge0-120B-q3k passed 27/50 and 21/50 harder synthetic coding attempts in separate live-context runs without execution errors; earlier runs encountered OOM responses and prefill timeouts. Separate CLI accounting passed 21/21 direct, 36/36 shadow and 15/15 retry final checks; two retry tasks needed frontier repair. The retry showed no speed win or promotion conclusion. Broader comparisons are outside this selected release.
What could this mean for your business?
Explore Practical AIPlain-language guides to choosing a task, checking the costs, and testing local AI before spending. Each guide connects practical questions to the research and explains what still needs testing in your own workflow.
Start with what local AI can do for a small businessTechnical write-ups.
Finding and Fixing Silent Data Loss in a Live Lead System
A system that collects about 6,400 businesses twice a day, opened with a formal check of its own data quality. It found two serious problems that never raised an error: one quietly merged 545 different businesses into others, and another threw away 57.8% of one collector's results. I then rebuilt it with 197+ automated checks so the same kind of bug can't hide again.
Getting Found and Used by AI Assistants: A Test on Porsync.com
A May 2026 snapshot: porsync.com was tested against two companies that sell tools for exactly this. The same AI commerce assistant was run once on each of the three sites. It succeeded on porsync.com and failed on the other two. The site has changed since, and a rerun is pending.
Updating a Site's WebMCP Setup for the Current Browser API: What Changed and What I Changed
WebMCP, the browser API that lets a page offer tools to AI agents, has moved since porsync.com first used it in May. I read Chrome's current documentation and the early adopters' write-ups, then updated this site: tools now register with a clean-up signal, carry read-only labels, and a second tool looks up a single research write-up. Tested against a production build with a stand-in for the browser API, not yet in Chrome with the real one.