Writing
Technical analysis and methodology from production ML, credit risk, and applied LLM work.
- Articles
- 7
- Published
- April to July 2026
- Words
- 21k
All articles
-
Self Bootstrap Exfiltration in Open Weights Agents: Absent Without Instruction, Induced by Injection
20+ configurations across 9 families (12 to 120B)
July 2026
-
I Built a System to Map My Own Beliefs. It Showed Me the Easy Half of a Harder Problem.
Belief extraction asymmetry: what gets surfaced, and what operates invisibly.
June 2026
-
The Refusal Axis Is Layer Local: Why Single Vector Steering Hits a Geometric Ceiling
Models: Qwen3.5-9B and Mistral-7B
June 2026
-
We Cut KYC From 7 Days to 30 Min Using VLMs. The Hard Part Wasn't the Model.
Nigerian addresses are descriptive, not structured.
June 2026
-
Your LLM Extraction Pipeline Is Out of MRM Scope. Your Credit Model Isn’t. Now What?
SR 26-2 carved GenAI out of MRM scope.
May 2026
-
We Finetuned Qwen 3 32B to Replace an LLM API. The Evaluation Was Harder Than the Training.
How to evaluate a finetuned LLM for structured extraction when standard metrics don’t work.
May 2026
-
Your Best Feature Might Be Your Biggest Liability
Pre-deployment stability analysis for credit risk models.
April 2026