使用说明

Job Application Assistant

📁 Part of my portfolio: see this project and more →

Multi-agent AI system for job search, resume tailoring, and application tracking. Built with LangGraph orchestration and MCP (Model Context Protocol) server.

Live Demo

🚀 Try it live (after deployment)

Architecture

┌─────────────────────────────────────────────────────────────────┐
│                      Next.js Dashboard                          │
│   Dashboard  │  Job Search  │  Kanban Board  │  Resume Upload   │
└─────────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────────┐
│                       FastAPI Backend                            │
│   /api/jobs  │  /api/applications  │  /api/resumes  │  /api/agent│
└─────────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────────┐
│                  LangGraph Orchestrator                          │
│  ┌───────────────────────────────────────────────────────────┐  │
│  │                   Supervisor Agent                         │  │
│  └───────────────────────────────────────────────────────────┘  │
│         │                    │                    │              │
│         ▼                    ▼                    ▼              │
│  ┌───────────┐       ┌───────────┐       ┌───────────┐          │
│  │  Matcher  │       │  Tailor   │       │  Tracker  │          │
│  │   Agent   │       │   Agent   │       │   Agent   │          │
│  └───────────┘       └───────────┘       └───────────┘          │
└─────────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────────┐
│                        MCP Server                                │
│  search_jobs │ parse_resume │ generate_cover_letter │ CRUD ops  │
└─────────────────────────────────────────────────────────────────┘
                              │
                              ▼
            ┌─────────────────────────────────────┐
            │  Adzuna API  │  RemoteOK  │  SQLite │
            └─────────────────────────────────────┘

Features

Job Search: Search Adzuna and RemoteOK APIs, auto-save results
AI Matching: LLM-powered skill matching with explanations
Cover Letter Generation: Claude generates tailored cover letters
Resume Suggestions: AI analyzes job requirements vs your resume
Kanban Tracking: Drag-and-drop application pipeline
MCP Server: 10+ tools following Model Context Protocol

Tech Stack

Layer	Technology
Agent Framework	LangGraph
MCP Server	FastMCP (Python)
Backend	FastAPI
Frontend	Next.js 14, shadcn/ui, TanStack Query
Database	SQLite + SQLAlchemy
LLM	Claude (Anthropic)
Job APIs	Adzuna, RemoteOK

Quick Start

Prerequisites

Python 3.11+
Node.js 18+
Anthropic API key (optional, for AI features)

Setup

# Clone
git clone https://github.com/EiriniOr/job-application-assistant
cd job-application-assistant

# Backend
python -m venv .venv
source .venv/bin/activate
pip install -e .

# Set environment variables
cp .env.example .env
# Edit .env with your API keys

# Frontend
cd frontend
npm install
cd ..

# Run
uvicorn backend.main:app --reload &
cd frontend && npm run dev

Open http://localhost:3000

Deploy to Cloud

Option 1: Railway (one-click)

Go to railway.app
"Deploy from GitHub repo" → select job-application-assistant
Add env vars: ANTHROPIC_API_KEY, PORT=8000
Deploy — Railway auto-detects Dockerfile

Option 2: Render.com

Go to render.com
"New Blueprint" → connect GitHub → select repo
Uses render.yaml to configure services automatically

Option 3: Backend + Vercel Frontend

Deploy backend to Railway/Render
Go to vercel.com → import repo
Root directory: frontend
Add NEXT_PUBLIC_API_URL = backend URL

With Docker

docker-compose up

Agents

Supervisor

Routes requests to specialist agents based on action type.

Matcher Agent

Searches Adzuna + RemoteOK APIs
Calculates match scores using Claude
Identifies skill overlaps and gaps

Tailor Agent

Generates personalized cover letters
Suggests resume bullet improvements
Maps experience to job requirements

Tracker Agent

Updates application status
Generates pipeline summaries
Tracks application timeline

MCP Tools

Tool	Description
`search_jobs`	Search across job boards
`get_job_details`	Fetch full job info
`parse_and_save_resume`	Extract resume structure
`create_job_application`	Save job to pipeline
`update_application`	Change application status
`save_cover_letter`	Store generated letter
`save_match_score`	Record match analysis

API Endpoints

GET  /api/jobs           - List saved jobs
GET  /api/jobs/search    - Search job boards
POST /api/applications   - Create application
GET  /api/applications   - List applications
PATCH /api/applications/:id - Update status
POST /api/resumes/upload - Upload resume
POST /api/agent/run      - Trigger agent workflow

Environment Variables

ANTHROPIC_API_KEY=     # Required for AI features
ADZUNA_APP_ID=         # Optional: Adzuna job search
ADZUNA_APP_KEY=        # Optional: Adzuna job search
DATABASE_URL=          # Default: sqlite+aiosqlite:///./data/jobs.db

License

MIT