How snowxx456/neural-pilot works
Project Type & Purpose
• This is a full-stack AI/AutoML web application (“NEURO PILOT”) with:
– Next.js TypeScript frontend (UI for dataset search, preprocessing, visualization, model training)
– Django REST Framework backend (API endpoints under “api” app)
– Custom ML modules (in Backend/model) for data cleaning, visualization, training
– SQLite for persistence
– External services: Groq LLM API for PandasAI-powered cleaning, Kaggle API for dataset search
File Structure & Architectural Clues
• Backend/
– config/ (Django settings, URLs, WSGI)
– api/ (models, views, serializers, migrations)
– llm_clean_data_script.py (entry to LLM cleaning)
– model/ (data_cleaning/, data_visualization/, modeltraining/)
– media/ (uploaded datasets & persisted models)
– manage.py, .env.example, requirements.txt
• Frontend/
– app/ (Next.js “app router” pages: chatbot, preprocessing, visualization, model_training)
– components/ (UI primitives, dataset search, charts, theme providers)
– lib/ (api wrapper, types, utils)
– hooks/, config files (ESLint, Tailwind, TS)
Key README Insights
• CLI-based origins, extended to UI
• Core flows: dataset search → automatic preprocessing → visualization → model training (15+ algorithms)
• Groq LLM & PandasAI used for conversational cleaning & feature engineering