What I Measure Before Calling a Model Ready
Accuracy is only one part of deciding whether an applied machine-learning system is ready to ship.
Read articleWriting and field notes
Blog
Notes on Edge AI, IoT and building things.
Designing IoT Pipelines for Unreliable Connectivity
What changes when sensors and users cannot depend on a constant internet connection.
Read articleRunning Quantized Models on Jetson Orin Nano
A practical guide to taking a computer-vision model from training to low-latency edge inference.
Read article