🚀 Re:You – Your AI-Powered Developer Brain

Re:You is an AI-powered semantic search and code understanding engine that lets developers recall past implementations, retrieve code snippets, explore commit history, and understand features across repositories — all through natural language queries.

It acts as a long-term memory layer for developers and teams.

---

✨ Features

• 🔍 Semantic Search Across Repos — Search by meaning, not just keywords.
• 🧠 Contextual Q&A — Ask “How does login work?” and get structured code answers.
• 📄 Code Snippet Retrieval with file paths and metadata.
• 🕒 Commit Insights — Understand how features changed over time.
• 📚 Embeddings-Based Indexing of functions, classes, and commits.
• ⚡ RAG Pipeline using Groq LLM + Chroma vector store.

---

🏗️ Architecture Overview

1. Data & Ingestion Layer

• GitHub repo cloning (local for MVP)
• Code extraction using AST (Python) & regex (JS)
• Commit extraction using Git
• Chunking functions/classes with metadata

2. Storage & Retrieval Layer

• Embeddings via MiniLM (current)
• ChromaDB as vector store
• Metadata stored alongside chunks
• Hybrid retrieval & reranking (future)

3. Query-Answering Layer

• Retrieval-Augmented Generation (RAG)
• LLM: Groq API (DeepSeek LLaMA model)
• Structured answers with citations
• CLI demo interface (frontend coming soon)

---

🚀 Getting Started

✅ 1. Create & Activate a Virtual Environment

Create a new environment

python -m venv venv

Activate the environment (Git Bash)

source venv/Scripts/activate

If using Linux/macOS:

source venv/bin/activate

---

✅ 2. Install dependencies

Once your virtual environment is active:

pip install -r requirements.txt

---

✅ 3. Set API Keys

Create a .env file in the project root:

GROQ_API_KEY=your_key_here
GITHUB_ACCESS_TOKEN=your_github_access_token

---

✅ 4. Generate embeddings

python embeddings/store_embeddings.py

---

✅ 5. Run the QA service

python qa/qa_service.py

---

📌 Project Structure

devmemory/
│── extraction/          # Code + commit extraction
│── embeddings/          # Chunking + vector generation
│── qa/                  # RAG pipeline / answer generation
│── retrieval/           # Retrieval logic (vector search)
│── ingestion/           # Repo ingestion + parsing
│── vector_store/        # Auto-generated embeddings DB
│── data/repo/           # Your cloned GitHub repo
│── README.md
│── requirements.txt