How satvik314/educhain works
Project type and purpose
• Educhain is an open-source Python library (packaged via setup.py) that provides a high-level API for generating personalized educational content using Generative AI.
• It offers two core “engines” (QnA and Content) behind a single facade class (educhain/core/educhain.py), plus supporting utilities, Pydantic models, and integrations with third-party LLMs.
File structure
• Top-level:
• setup.py, README.md, LICENSE, CONTRIBUTING.md, .gitignore – packaging & metadata
• docs/ – static documentation (features, getting-started)
• cookbook/ – example Jupyter notebooks, starter-apps, use-cases to demonstrate usage
• educhain/ – core package
• archive/ – old experiments & prototypes
• educhain_llms.txt – list of supported LLM backends
• educhain/ sub-folders:
• core/
– config.py (LLMConfig, default settings)
– educhain.py (Facade class)
• engines/
– qna_engine.py (Question generation workflows)
– content_engine.py (Lesson plans, flashcards, notes)
• models/
– base_models.py, qna_models.py, content_models.py (Pydantic schemas)
• utils/
– loaders.py (PDF, URL, text, image, video fetchers)
– output_formatter.py (export to JSON/CSV/PDF)
README for additional insights
1. QnA Engine
2. Content Engine
3. Personalization Engine (future)
• Key features: multiple LLM support via LangChain connectors, customizable prompt templates, export functionality, multimodal inputs (YouTube, images).
Detailed, essential architectural elements
• Don’t just list files—group by function (e.g., core vs engines vs utils)
• Consumer (notebooks/apps) → library → LLM → response path
• Pydantic layer between raw LLM output and final data model
• Export module as a first-class component
• Versioned roadmap items (future Personalization Engine)