README.md

metadata

title: AI Coding Assistant
emoji: ⚡
colorFrom: blue
colorTo: purple
sdk: docker
app_port: 7860
pinned: false
license: mit

AI Coding Assistant

Production-grade RAG-based coding assistant with LangChain, FAISS, and LoRA-tuned LLMs.

Features

• Semantic Code Search with FAISS vector store
• LangChain RAG pipeline for context-aware responses
• DeepSeek Coder LLM with optional LoRA fine-tuning
• 8-bit quantization for efficient inference (GPU only)
• Optional Qdrant Cloud integration

Usage

1. Enter your code repository path (or use sample data)
2. Click "Index Repository" to process your codebase
3. Ask questions or request code fixes
4. Optionally enable LoRA-tuned model (requires GPU)

Configuration

For Qdrant Cloud integration, add secrets in Space settings:

• QDRANT_URL: Your cluster URL
• QDRANT_API_KEY: Your API key

Performance

• CPU inference: ~10-30s per response (free tier)
• GPU inference: ~2-5s per response (upgrade required)
• First load: ~2-3 minutes (model download)

Local Development

git clone https://github.com/Kash6/localCopilot
cd localCopilot

# Windows with GPU
setup_conda_gpu.bat
run_conda.bat

# Or use pip
pip install -r requirements.txt
streamlit run app.py

Architecture

• Vector Store: FAISS (default) or Qdrant Cloud (optional)
• Embeddings: all-MiniLM-L6-v2
• Reranker: ms-marco-MiniLM-L-6-v2
• LLM: DeepSeek Coder 1.3B
• Framework: LangChain + Streamlit