Flagship
HR Insight AI
3-service personal platform that predicts employee attrition and auto-generates executive reports for HR leaders.
title: HR Insight AI blurb: 3-service personal platform that predicts employee attrition and auto-generates executive reports for HR leaders. personas: [fullstack, data] flagship: true stack:
- TypeScript
- NestJS
- Next.js
- React 19
- Ant Design
- FastAPI
- Python
- XGBoost
- scikit-learn
- Prisma
- PostgreSQL
- Claude API
- Socket.IO
- JWT/RBAC links: github: https://github.com/SyrineLarbi
What it is
A 3-service platform (NestJS REST API + Next.js dashboard + FastAPI ML service) that predicts employee attrition risk and generates AI-written executive reports with action plans for HR leadership.
Highlights
- Multi-service architecture: NestJS API orchestrates a Python / FastAPI ML service and a Next.js / Ant Design dashboard, communicating over REST + WebSockets.
- Role-based access control with 4 roles (Admin / HR Manager / Team Manager / Viewer), JWT auth, Passport strategies, and granular team-level assignments via a join table.
- End-to-end ETL pipeline (Extract → Clean → Transform → Validate) in Python / pandas for ingesting employee CSV data into Postgres and preparing features for inference.
- XGBoost attrition classifier on engineered features (tenure, engagement, overtime, promotion gap, training, absenteeism); served through batch and single-employee endpoints with risk levels (LOW / MEDIUM / HIGH).
- Claude API integration to auto-generate report summaries and structured action plans, persisted alongside historical
RiskSnapshotrecords for trend tracking. - 9 domain entities in Prisma with cascade deletes and an
AuditLogtable capturing every privileged action for compliance. - Responsive Ant Design dashboard with charts, real-time updates via Socket.IO, PDF export (pdfmake + file-saver), role-scoped report generation.
Stack at a glance
NestJS · Next.js · FastAPI · XGBoost · Claude API · Prisma · Postgres · Socket.IO.