Turns a services company’s attendance, projects and completed work into a performance score people can actually see.
A performance operating system for a services company: attendance, employees and departments, shifts and leave, projects and tasks — plus a points engine that turns completed work into a score people can actually see.
Source private
Industry
Professional Services
Role
Architect & Full-Stack Engineer
Type
SaaS Platform
Year
2026
Status
In progress
Web + Windows agent
Surfaces
Socket.IO
Realtime
Points engine
Scoring
Overview
MOPA HRMS is an HR platform built around a specific idea: attendance and task completion should roll into a visible score rather than sitting in separate reports nobody reads. A points engine converts completed work into a number employees and managers both see, which makes performance a running signal instead of an annual review artifact.
The platform covers attendance, employees and departments, shifts and leave, and projects and tasks. Three surfaces share one NestJS API — a Next.js web app for employees, managers and admins, and a Flutter Windows desktop agent that handles attendance and work-from-home activity tracking through a pure Dart FFI input counter.
Background work runs on BullMQ over Redis, with Socket.IO pushing live updates to the web app.
The problem
A services company had attendance in one system, tasks in another and performance nowhere — reviews were assembled by hand once a year from data nobody had looked at since it was recorded. Work-from-home activity was entirely unmeasured.
The solution
One platform where attendance, shifts, leave, projects and tasks feed a points engine that produces a score visible to the person earning it, backed by a Windows desktop agent that records attendance and WFH activity without manual entry.
Technical stack
Frontend
Next.js 15
Tailwind CSS 4
TanStack Query
Flutter
Backend
NestJS 11
Socket.IO
Data
MySQL 8
Redis
Also used
Prisma 6
BullMQ
Dart FFI
System architecture
1
NestJS 11 API over Prisma 6 and MySQL 8 with JWT authentication
2
Redis with BullMQ for background job processing and Socket.IO for realtime push
3
Next.js 15 App Router web app with TanStack Query and Radix components
4
Flutter Windows desktop agent packaged as MSIX for attendance and activity capture
5
A standalone pure-Dart FFI input-counting package with no HRMS concepts, reusable on its own
Engineering challenges
Challenge
Measuring work-from-home activity risks building surveillance rather than a performance signal, and coupling that measurement to HR logic would make it impossible to reason about.
How I solved it
Kept input counting in a standalone Dart FFI package that knows nothing about HRMS — it counts and reports, with all interpretation happening in the platform. The measurement layer stays auditable and independently testable.
Challenge
Performance data that only managers see does not change behaviour.
How I solved it
Built the points engine so the score is visible to the employee earning it in the same app, making performance a continuous visible signal rather than something surfaced once a year.
Core features
Attendance with shift and leave management
Employee and department directory
Projects and task tracking
Points engine converting completed work into a visible score
Windows desktop agent for attendance and WFH activity
Pure Dart FFI input counting, packaged as MSIX
Realtime updates over Socket.IO
Background jobs on BullMQ and Redis
Role separation for employees, managers and admins
JWT authentication
Results & impact
Attendance, task completion and work activity converge into one visible score, replacing a once-a-year manual assembly of data with a running performance signal that employees and managers read from the same place.
Screenshots
What I learned
Isolating a measurement primitive from the domain that interprets it keeps sensitive telemetry auditable — and makes it reusable.
A performance metric changes behaviour only when the person being measured can see it.
