数智人事 (HRM System) – AI‑enhanced Human‑Resources Management

What it is

A full‑stack HRM (Human Resources Management) platform built with Spring Boot 3.4 (Java 17) on the back‑end and Vue 2.6 + Element UI on the front‑end.
Besides the usual HR functions (org chart, employee lifecycle, attendance, leave, overtime, payroll, social‑insurance, RBAC, workflow), it embeds a RAG‑based knowledge base that can answer employee HR questions using large‑language‑model (LLM) services (DashScope or OpenAI compatible).

Why it matters

Shows how a traditional enterprise system can be upgraded with generative‑AI features without rewriting the whole product.
Demonstrates a clean separation of pure‑calculation domain logic (overtime, salary) that can be unit‑tested independently of the framework.
Provides a ready‑to‑run example of integrating Flowable 8.0 workflow engine, Spring Security JWT double‑token auth, and a vector store (PostgreSQL + pgvector) for semantic search.

Core Features

Category	Highlights
HR Core	Org/department management, employee profile, fine‑grained RBAC, attendance & clock‑in, leave & overtime requests, payroll & social‑insurance configuration.
Workflow	Flowable 8.0 native integration, separate MySQL DB for workflow data, listeners for side‑effects (e.g., leave balance updates).
AI‑Powered Q&A	RAG pipeline: document upload → MinIO storage → chunking → embedding (DashScope/OpenAI) → pgvector vector DB → semantic retrieval → LLM‑generated answer with employee context (department, role).
Security	HttpOnly double‑token (access + refresh) stored in cookies, transparent refresh via Axios interceptor, XSS‑resistant, Spring Security RBAC.
Infrastructure	Multi‑engine persistence: MySQL (business), PostgreSQL + pgvector (vectors), Redis (cache & captcha), MinIO (object storage).
Observability	Swagger‑UI (OpenAPI 3.1) auto‑generated docs, ECharts dashboards for attendance analytics.

Architecture at a Glance

graph TB
  subgraph Front[Vue 2.6 + Element UI]
    UI[UI components] --> Router[Vue‑Router (dynamic menus)]
    Router --> Store[Vuex (user, token, permissions)]
    Store --> Axios[Axios (double‑token refresh)]
  end
  subgraph Auth[Spring Security]
    JWT[JwtAuthenticationFilter] --> RBAC[RBAC checks]
  end
  subgraph Backend[Spring Boot 3.4]
    Service[Application services] --> Domain[Pure‑calc modules (OvertimeCalculator, SalaryCalculation)]
    Service --> Flowable[Flowable engine (MySQL flowable DB)]
    Service --> Knowledge[KnowledgeService (ETL, vector store)]
    Service --> Chat[AI Assistant (LLM & embedding)]
  end
  subgraph Storage[Persistence]
    MySQL[(MySQL – business)]
    FlowDB[(MySQL – flowable)]
    PG[(PostgreSQL + pgvector – vectors)]
    Redis[(Redis – cache/captcha)]
    MinIO[(MinIO – files)]
  end
  Front -->|HTTP/REST| JWT --> Auth --> Service
  Service --> MySQL
  Service --> FlowDB
  Knowledge --> MinIO
  Knowledge --> PG
  Chat --> LLM[Configurable LLM]
  Chat -.retrieval.-> PG

Tech Stack

Layer	Tools & Versions
Language	Java 17, JavaScript (ES6)
Back‑end Framework	Spring Boot 3.4.4, Spring Security 6.x, MyBatis‑Plus 3.5.10
Workflow Engine	Flowable 8.0.0
AI Integration	Spring AI 1.0.0 (spring‑ai‑alibaba 1.0.0.2), compatible with DashScope or OpenAI APIs
Vector Store	PostgreSQL 16 + pgvector extension
Cache / Session	Redis 5.0
Object Storage	MinIO (S3‑compatible)
Front‑end	Vue 2.6.14, Element UI 2.15.7, Vue‑Router 3.2, Vuex 3.6, Axios 0.25, ECharts 5.3
Containerisation	Docker Compose (MySQL, Redis, PostgreSQL, MinIO)
Docs / API	SpringDoc OpenAPI 2.8 (Swagger UI)

Getting Started Locally

Start middleware
```
docker compose -f docker/local/docker-compose.yml up -d
```
(MySQL 8.1, Redis 5, PostgreSQL 16 with pgvector, MinIO)

Import DB schemas

mysql -h 127.0.0.1 -P 3306 -uroot -p123456 hrm < db/mysql/hrm.sql
mysql -h 127.0.0.1 -P 3306 -uroot -p123456 hrm_flowable < db/mysql/hrm_flowable.sql
psql -h 127.0.0.1 -p 5432 -U hrm -d hrm_kb -f db/postgresql/knowledge_base.sql

Configure AI credentials (optional for the smart Q&A)

# create hrm-server/.env (see .env.example)
LLM_API_KEY="your‑key"
JWT_SECRET="random‑secret"
LLM_BASE_URL="https://dashscope.aliyuncs.com/compatible-mode"
LLM_CHAT_MODEL="qwen3.7-flash"
LLM_EMBEDDING_MODEL="qwen3.7-text-embedding"

Run the back‑end
```
cd hrm-server
mvn spring-boot:run   # starts on port 8888
```
Swagger UI: http://localhost:8888/swagger-ui.html

Run the front‑end

cd hrm-admin
npm ci
npm run serve   # starts on port 8080

Login
- URL: http://localhost:8080
- User: admin
- Password: 123456
- Captcha is stored in Redis; the UI fetches it automatically.

Typical Use Cases

Employee self‑service – view personal profile, submit leave/overtime, check payroll, query HR policies via the AI chat.
HR admin – manage org structure, configure RBAC, define approval processes, upload policy documents that become searchable by the RAG engine.
Manager – approve/reject requests, see attendance analytics, receive notifications generated by workflow listeners.
AI‑enhanced support – an employee asks “How many days of annual leave do I have left?” → system retrieves the relevant policy snippet, combines it with the employee’s leave balance, and the LLM returns a concise answer.

Who Might Benefit

SMEs & startups looking for a ready‑made, AI‑augmented HR platform that can be self‑hosted.
Enterprises that want to experiment with RAG‑based internal knowledge bases without building the pipeline from scratch.
Developers interested in seeing a real‑world integration of Flowable, Spring AI, and pgvector within a monolithic Spring Boot app.

License & Contribution

The README does not specify a license, so check the repository root for a LICENSE file before commercial use. Contributions appear welcome; the project includes a docs/ folder with development specs and test reports, and the back‑end is organized by domain, making it easy to add new modules or swap the LLM provider via configuration.

Bottom line: 数智人事 is a genuine, production‑grade HRM system that showcases how generative AI (RAG) can be layered on top of classic enterprise software, offering a concrete reference implementation for developers building AI‑enhanced business applications.

quuuuj/hrm