Document measured API p95 latency in README

Remove unused shap dependency

Auto-restart FastAPI backend under supervisord if it crashes

Disclose recommendation-diversity finding and AMD calibration-set shortfall

Serve FastAPI and Streamlit together; UI calls API over HTTP

Close power-efficiency Pareto gap, add training-data disclosure, harden data lineage, cut Docker image 84%

Add MIT license, refresh README accuracy numbers post-calibration

GPU Perf Prophet v1 — cross-vendor GPU performance forecasting & recommendation engine