Skip to content

README.Project-Structure.md

Latest commit

 

History

History
129 lines (105 loc) · 4.03 KB

README.Project-Structure.md

File metadata and controls

129 lines (105 loc) · 4.03 KB

Generative AI Project Template

A structured project template with best practices for private-first and scalable Generative AI applications.

🌟 Features

  • Modular project structure for scalability
  • Pre-configured support for multiple LLM providers (Claude, GPT)
  • Built-in prompt engineering utilities
  • Rate limiting and token management
  • Robust error handling
  • Caching mechanism for API responses
  • Example implementations and notebooks

📁 Project Structure

generative_ai_project/
├── config/                   ## Configuration directory
│   ├── __init__.py
│   ├── model_config.yaml     ## Model-specific configurations
│   ├── prompt_templates.yaml ## Prompt templates
│   └── logging_config.yaml   ## Logging settings
│
├── src/                      ## Source code
│   ├── llm/                  ## LLM clients
│   │   ├── base.py           ## Base LLM client
│   │   ├── claude_client.py  ## Anthropic Claude client
│   │   ├── gpt_client.py     ## OpenAI GPT client
│   │   └── utils.py          ## Shared utilities
│   │
│   ├── prompt_engineering/   ## Prompt engineering tools
│   │   ├── templates.py      ## Template management
│   │   ├── few_shot.py       ## Few-shot prompt utilities
│   │   └── chain.py          ## Prompt chaining logic
│   │
│   ├── utils/                ## Utility functions
│   │   ├── rate_limiter.py   ## API rate limiting
│   │   ├── token_counter.py  ## Token counting
│   │   ├── cache.py          ## Response caching
│   │   └── logger.py         ## Logging utilities
│   │
│   └── handlers/             ## Error handling
│       └── error_handler.py
│
├── data/                     ## Data directory
│   ├── cache/                ## Cache storage
│   ├── prompts/              ## Prompt storage
│   ├── outputs/              ## Output storage
│   └── embeddings/           ## Embedding storage
│
├── examples/                 ## Example implementations
│   ├── basic_completion.py
│   ├── chat_session.py
│   └── chain_prompts.py
│
└── notebooks/                ## Jupyter notebooks
    ├── prompt_testing.ipynb
    ├── response_analysis.ipynb
    └── model_experimentation.ipynb

🚀 Getting Started

  1. Clone the repository:
git clone https://github.com/nnthanh101/ai-engineering.git
cd ai-engineering
  1. Install dependencies:
pip install -r requirements.txt

  1. Configure your environment:

    • Copy config/model_config.yaml.example to config/model_config.yaml
    • Add your API keys and configurations

  2. Review the examples in examples/ directory

  3. Start with the notebooks in notebooks/ for experimentation

📘 Documentation

Configuration

  • model_config.yaml: Configure API keys and model parameters
  • prompt_templates.yaml: Define reusable prompt templates
  • logging_config.yaml: Configure logging behavior

Key Components

  1. LLM Clients (src/llm/)

    • Base client with common functionality
    • Specific implementations for different providers
    • Utility functions for token counting and rate limiting

  2. Prompt Engineering (src/prompt_engineering/)

    • Template management system
    • Few-shot prompt utilities
    • Prompt chaining capabilities

  3. Utilities (src/utils/)

    • Rate limiting for API calls
    • Token counting
    • Response caching
    • Logging

🛠️ Development

Best Practices

  • 1. Keep configuration in YAML files
  • 2. Implement proper error handling
  • 3. Use rate limiting for APIs
  • 4. Maintain separation between model clients
  • 5. Cache results when appropriate
  • 6. Document your code
  • 7. Use notebooks for experimentation

Tips

  • Follow modular design principles
  • Write tests for new components
  • Use proper version control
  • Keep documentation updated
  • Monitor API usage and limits