CV Maker

AI-powered CV generator that creates tailored, Harvard-style, ATS-compatible CVs from your persistent background memory.

Features

• Persistent memory — stores your background (education, experience, projects, skills) in memory.json
• AI generation — uses Groq (llama-3.3-70b-versatile) to write tailored CV content
• Research mode — automatically searches the web for application requirements when you name a target (e.g. "ETH Zurich MSc CS 2025")
• Bilingual — outputs CVs in English or Turkish
• Harvard style — standard academic CV format, ATS-compatible
• PDF export — downloads a ready-to-send PDF

Setup

1. Install Python 3.11+

2. Install dependencies:

   pip install -r requirements.txt

3. Copy .env.example to .env and fill in your API keys:

   cp .env.example .env

   • GROQ_API_KEY — required. Get from console.groq.com (free)
   • TAVILY_API_KEY — optional. Get from app.tavily.com (free tier: 1000 searches/month). If omitted, DuckDuckGo is used for web research.

4. Start the app:

   uvicorn app.main:app --reload

5. Open http://localhost:8000 in your browser.

Usage

Build your memory

Go to My Memory and add your background:

• Paste text (work history, project descriptions, bio)
• Upload a PDF, DOCX, or TXT resume/CV

The AI extracts and merges the information into memory.json.

Generate a CV

Go to Generate CV and enter a target:

• Paste a job description for a direct match
• Type a target name (e.g. "Google SWE internship", "MIT PhD CS 2026") — the agent will research requirements automatically

Select your output language (English or Turkish) and click Generate CV. Download the PDF when it appears.

Running tests

pytest -v

Notes

• WeasyPrint (high-quality HTML→PDF) requires GTK on Windows. If not installed, the app automatically falls back to fpdf2 for PDF generation.
• The app runs locally and does not send data anywhere except to Groq API (for CV generation) and Tavily/DuckDuckGo (for web research when needed).
• memory.json is gitignored — your personal data stays on your machine.