RAG Hybrid Search Engine

A semantic search engine built with C#, Ollama, and Qdrant Cloud.

What it does

• Loads any PDF and splits it into chunks
• Converts chunks to vectors using Ollama (nomic-embed-text)
• Stores vectors in Qdrant Cloud vector database
• Searches using Hybrid Search - Vector similarity + BM25 keyword search combined
• Results ranked with Reciprocal Rank Fusion (RRF)

Tech Stack

• Language: C# / .NET 8
• Embeddings: Ollama (nomic-embed-text) - runs locally, free
• Vector Database: Qdrant Cloud
• PDF Parsing: PdfPig
• Search: Hybrid (Vector + BM25)

Setup

Requirements

• .NET 8 SDK
• Ollama installed and running
• Qdrant Cloud account (free tier works)

Installation

1. Clone the repo git clone https://github.com/yourusername/QdrantRAG.git cd QdrantRAG

2. Create .env file with your keys QDRANT_URL=your_qdrant_url_here QDRANT_API_KEY=your_qdrant_api_key_here

3. Pull the embedding model ollama pull nomic-embed-text

4. Run dotnet run

How it works

1. PDF text is extracted page by page
2. Text is cleaned (spaces fixed) and split into 40-word overlapping chunks
3. Each chunk is embedded into a 768-dimensional vector
4. Vectors stored in Qdrant with original text as payload
5. At query time - both vector search AND BM25 run in parallel
6. Results merged with Reciprocal Rank Fusion for best accuracy

30-day AI Engineering sprint

Part of a 30-day plan to align with AI/LLM engineering role requirements.