Tools & methods
The tools I use across vision experiments, services and delivery, with the project where each appears. Where the code is public I link it; where it is not, I describe the use in bounded terms. Depth varies by project.
Vision and deep learning
PythonPyTorchDINOv2
A DINOv2 ViT-B/14 backbone feeding four linear classification heads in Fashion Attribute Recognition.
Stable-Baselines3Gymnasiumhighway-env
PPO training and checkpoint evaluation in simulated highway driving in Highway RL Agent.
Backend and data
FastAPIPydanticUvicorn
An inference service with upload validation, health endpoints and an API contract in Fashion Attribute Recognition.
Frontend
Next.jsReactTypeScriptTailwind CSS
The stack this website runs on, built with the help of AI coding assistants. None of the project repositories above contains an interface.
HTML5 Canvas
The Sobel-edge viewer on the Fashion page. It is illustrative and does not run the classifier.
Environments and delivery
DockerDocker Compose
A container image and an example Compose file for the Fashion service.
GitHub Actionspytestruffmypy
A CI workflow with tests, lint and type checks for Fashion Attribute Recognition.