Case study · 2026
Check-in Med staff check-in by face in Telegram
No fingerprint terminal and no cards to pass around: an employee opens the bot, looks into the camera and is checked in within seconds. The phone recognises the face, and geolocation confirms they are at the branch.
- Role
- Product and code
- Stack
- PHP 8, Preact, SQLite or MySQL
- Face recognition
- On the phone, WebAssembly
- Interface
- Russian and Uzbek
Full-stack developer
Clinic staff check in with a face scan inside Telegram; managers see who came, who was late and the monthly rating.
How a check-in works
- The employee opens the bot — the face scanner starts right away.
- The phone computes a face print and checks that it is a live person, not a photo.
- Geolocation confirms the employee is inside the branch radius.
- “You're checked in!” — and the app closes itself.
For managers
- Today at a glance — scheduled, arrived, late, absent, left early.
- Monthly rating by lateness, hours and absences, with Excel export.
- Employee calendar — every day colored: on time, late, absent, on leave.
- Branches on the map with a check-in radius from 50 to 500 m.
- Any schedule — weekly, rotating, night and 24-hour shifts, holidays, leave and an action log.
Under the hood
Plain PHP 8 without a framework, so it runs on ordinary shared hosting. The heavy face recognition happens on the phone; the server only compares face prints, applies the rules and stores data. Face templates are encrypted with AES-256-GCM, and 70 automated scenarios cover check-ins, schedules and reports. The screens here show the app's built-in demo data.
Start a project
Let's build it.
Describe the task in a couple of sentences. I'll reply with questions, a plan and a fixed price — usually the same day.