My Assistant API

A TypeScript-based API service that powers an AI assistant capable of processing documents, handling profile information, and providing intelligent search capabilities.

Features

• 🤖 AI-powered document processing and analysis
• 📝 Profile ingestion and management
• 🔍 Vector-based semantic search using Pinecone
• 📚 Document chunking and processing utilities
• ⚡ Built with TypeScript and modern cloud technologies

Tech Stack

• Language: TypeScript
• AI Integration: LangChain, OpenAI
• Vector Store: Pinecone
• Development:
  • Vitest for testing
  • ESLint for code quality
  • Cloudflare Workers for deployment

Project Structure

my-assistant-api/
├── src/                    # Source code
│   ├── handlers/          # Request handlers
│   ├── libs/              # Core libraries
│   │   ├── ai/           # AI/LLM integrations
│   │   ├── document/     # Document processing
│   │   ├── kv/           # Key-value storage
│   │   └── vectorstore/  # Vector database integration
│   └── types.ts          # Type definitions
├── test/                  # Test files
└── assets/               # Static assets and documents

Getting Started

1. Install dependencies:

   npm install

2. Set up environment variables:

   • Configure your Cloudflare Worker secrets
   • Set up OpenAI API keys
   • Configure Pinecone environment

3. Run tests:

   npm test

4. Start development:

   npm run dev

Key Components

• Document Processing: Handles document chunking, summarization, and metadata extraction
• Profile Ingestion: Manages user profile data and related information
• Search: Implements semantic search capabilities using vector embeddings
• AI Integration: Provides LLM-powered features through OpenAI and LangChain

Development

The project follows TypeScript best practices and includes:

• Strong typing with TypeScript
• Unit tests with Vitest
• ESLint configuration for code quality
• Cloudflare Workers for serverless deployment

License

MIT