I need a web-based face recognition system that records employee attendance the moment someone shows up, without any manual check-in. Accuracy and speed are more important to me than flashy UI, but the interface must still be clear enough for HR staff to view logs and pull reports with zero training. Core functionality I expect: • Real-time attendance logging as soon as a face is detected • Instant notification alerts (e-mail, push, or SMS—whatever integrates cleanly) when someone signs in, signs out, or when an unknown face appears • Report generation that can be filtered by date range, department, or individual employee, exportable to CSV and PDF • A webhook endpoint so our internal systems can consume every check-in or alert programmatically Feel free to choose the stack you are most productive with—Python (OpenCV, FaceNet, Flask/FastAPI), or a JS stack (TensorFlow.js, Node, React) are both fine—as long as deployment is straightforward on a standard cloud VM or Docker container. The model must run on-prem or in our private cloud; no third-party APIs that send images off-site. Acceptance criteria: 1. System recognises registered employees with 98 %+ accuracy under normal office lighting. 2. Dashboard updates within two seconds after a scan. 3. Reports match the raw log data exactly; spot checks will be performed. 4. Webhook fires a POST with JSON payload for every check-in/out event. 5. Complete source code, Dockerfile, and a short readme that lets us deploy from scratch. If you have built something similar, link a short demo or repo so I can assess fit quickly.