Qdrant Go POC: Semantic RAG & Vector Search

A high-performance "showoff" project demonstrating the power of Qdrant, Go, and Google Gemini for building production-grade Retrieval-Augmented Generation (RAG) systems.

🚀 Overview

This application provides a real-time interface to interact with a semantic knowledge base. It's built for speed, transparency, and persistence, using a modern Go-based stack.

Core Tech Stack:

• Qdrant: High-performance vector database for similarity search and metadata filtering.
• Google Gemini: State-of-the-art text-embedding-004 and gemini-1.5-flash for generative RAG.
• Go + HTMX + Templ: A "Hypermedia-driven" approach for a reactive UI without the complexity of a JS framework.
• SQLite: Local persistence for chat history and metadata.

🛠 Architecture

graph TD
    User((User)) -->|Question| UI[HTMX/Templ UI]
    UI -->|Multipart Upload| Backend
    UI -->|Chat Request| Backend[Go Backend]
    
    Backend -->|History + Context| SQLite[(SQLite History)]
    Backend -->|Text| GeminiEmbed[Gemini Embedding API]
    GeminiEmbed -->|Vector| Backend
    
    Backend -->|Query Vector| Qdrant[Qdrant Vector DB]
    Qdrant -->|Context + Scores| Backend
    
    Backend -->|Memory + Context| GeminiLLM[Gemini 1.5 Flash]
    GeminiLLM -->|Streamed Answer| Backend
    
    Backend -->|HTMX Update| UI
    Backend -->|Action Logs| UI

Loading

✨ Advanced Features

• Dynamic Knowledge Management: Upload .txt or .md files directly via the UI. Files are automatically chunked, embedded, and indexed.
• Conversational Memory: The assistant maintains context of the last 5 messages for natural follow-up questions.
• Streaming Responses: Real-time character-by-character generation for a responsive user experience.
• Source Citations: Every answer includes the original document sources and their similarity scores.
• Persistent History: Conversations are saved to a local SQLite database and reloaded on restart.
• Real-time Observability: Watch every step of the RAG pipeline (Embedding -> Retrieval -> Generation) in the live "Action Logs".

💡 Vector DB Use Cases

Beyond simple RAG, Qdrant and embeddings enable:

1. Semantic Search: Find documents by meaning, not just keywords.
2. Product Recommendations: Suggest items based on visual or textual similarity.
3. Anomaly Detection: Identify outliers in high-dimensional space for fraud or monitoring.
4. De-duplication: Find near-identical entries in massive datasets where exact hash matching fails.

🚦 Getting Started

1. Prerequisites

• Docker & Docker Compose
• Google Gemini API Key

2. Setup

# Clone the repository
git clone https://github.com/your-username/qdrant-poc-go
cd qdrant-poc-go

# Set up environment variables
cp .env.example .env
# Edit .env and add your GEMINI_API_KEY

3. Run with Docker (Recommended)

The entire stack is containerized for easy deployment.

docker-compose up --build

4. Local Development

If you prefer running the Go app natively:

# Start only Qdrant
docker-compose up -d qdrant

# Run the Go application
go run cmd/app/main.go

Visit http://localhost:8080 to interact with the dashboard!

📸 Screenshots

(Add your screenshots here after running the app)

---

Developed as a high-signal POC for modern AI-driven applications.