How projectultra/employai works
Project type and purpose
• This is a full-stack Flask web application (“EmployAI”) that helps users generate CVs, cover letters, interview questions, and related materials using Python scripts and embedded LLM/quantization experiments.
• The README points to quantization and finetuning notebooks for LLMs (e.g., Gemma), so the core value-add is on-the-fly AI-powered document and question generation.
File structure
Top-level:
• app.py – Flask application entry point
• cover_letter.py – standalone cover-letter generator
• interview_questions.py – interview question generator
• df.csv – seed data (e.g., question banks, templates)
• requirements.txt – Python dependencies
• static/ – JavaScript, CSS, assets for client side
• templates/ – Jinja2 HTML templates for each view
• gemma_test/ – Jupyter notebooks and experiments (not part of production)
• *.ipynb – exploratory notebooks (e.g., interview_gen_ai.ipynb)
Architectural hints:
• MVC pattern—Flask routes (controllers) in app.py invoke Python modules (models), render Jinja2 templates (views).
• Static assets served from “static” directory; dynamic pages use “templates.”
• No dedicated database; a CSV is used as a read-only data source.
README for additional insights
• README provides links to quantization/finetuning of LLMs (Gemma) and indicates that the project is intended to tie into custom-quantized LLM inference for generation tasks.