AI / Computer Vision × Time & Attendance

Vision AttendanceGesture & Face Recognition Timeclock

A touchless time-and-attendance system where a gesture selects the intended punch type, face recognition matches the employee, and the resulting event is saved to the employee's daily timecard. The system is currently deployed across multiple skilled nursing facilities in the United States.

Next.jsNode.jsFirebasePythonKotlinSwift

THE PROBLEM

Turn gesture and face-recognition signals into a clear, reliable attendance flow that captures the intended punch and records it against the correct employee's daily timecard.

HARVEY'S ROLE

Cross-platform product engineering spanning a Next.js web UI, Node.js APIs, Firebase services, Python face recognition, Kotlin for Android, Swift for iOS, and a production deployment serving multiple U.S. skilled nursing facilities.

Claims are intentionally conservative and based only on supplied information.

INTERACTIVE ARCHITECTURE SIMULATION

Gesture, face match & timecard flow

GESTURECapture user intent
PUNCH TYPERequested action
FACE RECOGNITIONCapture and compare
EMPLOYEE MATCHIdentity confirmed
DAILY TIMECARDReady to save
VISION WORKFLOW / REPRESENTATIVE DEMO

Punch & Timecard Flow

  1. 1Gesture detected
  2. 2Punch type selected
  3. 3Face captured
  4. 4Employee matched
  5. 5Attendance event validated
  6. 6Daily timecard updated
  7. 7Confirmation displayed
REPRESENTATIVE ATTENDANCE EVENT

Daily timecard

CLOCK INPUNCH TYPE

Waiting for gesture and face match

AWAITING RUN

SYSTEM ARCHITECTURE

A readable path from interface to system.

Conceptual architecture based only on verified technologies.

01Camera
02Gesture Recognition
03Punch-Type Decision
04Face Recognition
05Employee Match
06Daily Timecard

ENGINEERING DECISIONS

Make every recognition step explicit.

Use gesture recognition first to determine the user's intended punch type.

Run face recognition as a separate employee-matching step before saving attendance.

Make gesture, match, validation, and timecard-save states visible in the interface.

TRADEOFFS & INTEGRITY

Real workflow. Private employee data.

The public portfolio demo uses a representative employee identifier and confidence value.

Production recognition accuracy, employee data, and attendance records are not disclosed.

PRODUCTION READINESS

Built for reliable production attendance.

The system separates gesture intent, face recognition, employee matching, and timecard updates into clear production-ready stages.

Designed for maintainability, clear validation boundaries, and consistent attendance workflows across platforms.

PROJECT MEDIA

Vision Attendance media archive

Project photos and videos are loaded directly from this case study's public media folder.

02 MEDIA FILES

TECHNOLOGY STACK

Cross-platform attendance engineering.

One production system spanning web, cloud services, computer vision, Android, and iOS.

01 · WEB UINext.jsResponsive employee and attendance interfaces
02 · APINode.jsApplication APIs and timecard workflows
03 · CLOUDFirebaseManaged application and data services
04 · AI / VISIONPythonFace-recognition processing and employee matching
05 · ANDROIDKotlinNative Android application
06 · iOSSwiftNative iOS application