AI Engineering Hub

Overview#

AI Engineering Hub is a large-scale open-source collection of tutorials and code examples, organized as a Monorepo containing dozens of independent sub-projects. It aims to bridge the gap between AI theory and engineering practice through runnable code and detailed Jupyter Notebooks.

Core Content Areas#

LLM Application Development#

• Local ChatGPT clones and chat interfaces
• Reasoning visualization dialogue systems
• Multi-model comparison and switching (DeepSeek, Llama, Qwen, OpenAI, Claude)

RAG Systems#

• Basic RAG workflows
• Agentic RAG with Web search fallback
• Multimodal RAG
• Vector database integration (Qdrant, Milvus, ColBERT)

AI Agents#

• Single-agent tool calling
• Multi-agent collaboration systems
• Automated workflows (book writing, brand monitoring, stock analysis)

MCP (Model Context Protocol)#

• Custom MCP Server implementation
• Cursor IDE tool integration
• Vector database connection and search

Model Fine-tuning & Evaluation#

• DeepSeek model fine-tuning (Unsloth + Ollama)
• Reasoning model building
• Model performance evaluation frameworks

Project Difficulty Levels#

Supported Model Ecosystem#

• Open Source: DeepSeek (R1/V3), Llama (3.2/4), Qwen (2.5VL/3), Gemma 3
• Commercial APIs: OpenAI (GPT/O3/O4), Anthropic Claude Sonnet 4, NVIDIA NIM

Quick Start#

# Clone repository
git clone https://github.com/patchy631/ai-engineering-hub.git
cd ai-engineering-hub/agentic_rag

# Install dependencies
pip install crewai crewai-tools qdrant-client fastembed

# Configure environment
cp .env.example .env

# Run application
streamlit run app_deep_seek.py

Typical Project Structure#

project_name/
├── assets/           # Static resources
├── knowledge/        # Knowledge base documents
├── src/              # Source code
├── .env.example      # Environment template
├── app.py            # Main application entry
├── *.ipynb           # Jupyter Notebook tutorials
└── README.md         # Documentation

Featured Sub-projects#

• agentic_rag: Document RAG with Web search fallback (CrewAI + DeepSeek-R1 + Qdrant)
• DeepSeek-finetuning: DeepSeek model fine-tuning tutorial (Unsloth + Ollama)
• mcp-agentic-rag: MCP-powered Agentic RAG (Bright Data + Cursor IDE)
• notebook-lm-clone: Full NotebookLM clone (RAG + citations + podcast generation)
• ai-engineering-roadmap: Complete AI engineer learning path

Use Cases#

• Self-learning & training: Master LLM engineering skills
• Rapid prototyping: Reuse RAG pipelines, Chat UIs, Agent workflows
• Technology validation: Compare actual performance across multiple models