Blog
Notes on agents, computer vision and AI engineering in production.
Agentic workflow v2: why I rebuilt it and how many errors it stopped
An audit of my own agent usage showed where the workflow was failing. V2 reviews the plan and the code on every task and runs in waves. Here are the errors it stopped, how I counted them and what it cost.
9 min read
Taking Python out of inference: a compiled YOLOv10s and my own C++ engine
I compiled the same network to AOTInductor, TensorRT, ONNX Runtime, OpenVINO and ncnn, then wrote a CPU engine from scratch with no inference library. I measured latency, dependencies and whether the outputs stayed the same.
10 min read
Claude plans, GPT executes: my agentic workflow
One Claude session thinks, decides and reviews. Execution goes to GPT agents isolated in worktrees. This is the workflow I use across my projects, from request to cherry-pick.
6 min read