job-hunting-assistant

# Job hunting assistant This is a AI assitant project implemented by python using LLM and relative technics, which aims to help me to find a job. ## Features - Web crawcle to collect the job descriptions published on the internet. - According to the market investigation and requirements, optimizing the resume to match the most needed position ## Key Tech Highlight - LLM's superpower to optimizing text structure - RAG statege to supply rich context for LLM - RAG summary content to limit tokens sent to LLM - MCP Server to provide the tools LLM need when to take actions ## System Architecture ```mermaid graph TB subgraph "Entry Point" A[__init__.py] --> B[server.py] end subgraph "Core Server" B --> C[JobSearchMCPServer] C --> D[FastMCP Instance] C --> E[Logger Configuration] end subgraph "Tools Layer" F[JobTools Class] --> G[get_joblist_by_expect_job] F --> H[get_job_by_resume] C --> F end subgraph "LLM Integration" I[LLMClient] --> J[OpenAI Client] J --> K[DeepSeek API] F -.-> I end subgraph "Web Scraping" L[listjob.py] --> M[Selenium WebDriver] M --> N[BOSS直聘网站] G --> L end subgraph "Prompt Management" O[prompt.py] --> P[Job_Search_Prompt Template] H --> O end subgraph "External Dependencies" N[BOSS直聘网站] K[DeepSeek API] Q[Local job.txt file] G --> Q end style A fill:#e1f5fe style C fill:#f3e5f5 style F fill:#e8f5e8 style I fill:#fff3e0 style L fill:#fce4ec ``` ## Data Flow - Client requests job recommendations via MCP protocol - Server retrieves job listings (currently from local file, can use web scraping) - Combines job data with user resume using AI prompt template - Sends to DeepSeek API for intelligent matching - Returns personalized job recommendations and advice ## How run the project - First, we need to install necessary packages - Then, prepare the data we need - Setup environment variables via the .env file - we use a local chromedriver to do web search, you should download it first chromedriver_path="/Users/fengshiyi/Downloads/chromedriver-mac-x64/chromedriver" - Now, we can use args to specify the working directory and run the server: ```bash uv --directory /Users/fengshiyi/Downloads/shayne/learning/LLM/py-projects/job-hunting-assistant/src/job_hunting_server run job-hunting-assistant ``` - Claude Desktop or Cline ```json { "mcpServers": { "job_hunting_server": { "command": "uv", "args": [ "--directory", "/Users/fengshiyi/Downloads/shayne/learning/LLM/py-projects/job-hunting-assistant/src/job_hunting_server", "run", "job-hunting-assistant" ] } } } ```