1. Create the Space (Gradio SDK, CPU basic is enough) at huggingface.co/new-space, or: huggingface-cli repo create face-expression-app --type space --space_sdk gradio 2. Upload these files to the Space root: app.py, matcher.py, extractor.py, requirements.txt, README.md (and examples/ if you made any). huggingface-cli upload 23f1000805/face-expression-app . . --repo-type space 3. If the model repo is private, add a Space secret HF_TOKEN. The first photo run downloads the Py-Feat weights, so it takes a minute.