📄 Semantic Search Engine (FastAPI + Embeddings)

A scalable Semantic Search Engine built with FastAPI that allows users to upload PDF documents, automatically extract and embed their contents, and perform semantic + metadata-aware search across stored documents.

The system follows a clean CSR (Controller–Service–Repository) architecture, supports tag-based filtering, and is designed to be extensible for multilingual embeddings.

---

Key Features

• PDF Upload

  • Upload PDF files via API
  • Automatic text extraction per page
  • Intelligent chunking for semantic indexing

• Semantic Search

  • Vector-based similarity search using embeddings
  • Natural language queries (not keyword-only)

• Tag Support

  • Assign multiple tags to PDFs (e.g. AI, ML, transformers)
  • Filter search results by tag

• Multi-Language Ready

  • Supports multilingual embedding models
  • Language stored as metadata per document

• Clean Architecture (CSR)

  • Controller layer (FastAPI routes)
  • Service layer (business logic)
  • Repository layer (data + vector DB)
  • Client layer (embedding models)

---

📁 Project Structure

text_embedding_system/
├── app
│   ├── main.py
│   ├── config.py
│   ├── models.py
│   ├── controllers
│   │   ├── entries.py
│   │   └── search.py
│   ├── services
│   │   ├── entry_service.py
│   │   └── search_service.py
│   ├── repository
│   │   └── dataset_repo.py
│   └── clients
│       ├── embedder_client.py
│       └── faiss_client.py
└── requirements.txt

---

🧩 Tech Stack

• Backend: FastAPI
• Language: Python 3.10+
• PDF Parsing: pypdf
• Vector Database: ChromaDB
• Embeddings: Sentence Transformers
• Validation: Pydantic
• Architecture: CSR Pattern

---

Author

Mostafa Abdelhamed