SEMBRA

Enterprise Document Intelligence Platform

Overview

SEMBRA is a high-performance document retrieval and intelligence platform that enables:

• Document Ingestion: Upload PDF, DOCX, TXT, Markdown files
• Intelligent Chunking: Configurable chunk size with overlap for context preservation
• Semantic Search: Hybrid vector + BM25 search with RRF fusion
• Graph Context: Document relationships and citation chains
• LLM Synthesis: AI-powered answers with source citations

Architecture

Component	Purpose	Port
Postgres	Auth, Config, Tenants	5432
BarqDB	Vector embeddings storage	8080
BarqGraphDB	Document graph relationships	8081
Celrix	High-speed embedding cache	6380
AiMesh	Message queue for embedding pipeline	9000
SEMBRA API	REST API	3000
SEMBRA UI	Web interface	80

Quick Start

# Clone repository
git clone https://github.com/YASSERRMD/sembra.git
cd sembra

# Start all services
cd docker && docker-compose up -d

# Access
# API: http://localhost:3000
# UI:  http://localhost

API Endpoints

Authentication

POST /v1/login
{"email": "admin@enterprise.com", "password": "password"}
# Returns: {"token": "eyJ..."}

Document Upload

POST /v1/upload
Content-Type: multipart/form-data
# files[]: PDF, DOCX, TXT, MD
# document_id: string
# document_name: string

Configure Embedding

POST /v1/configure-embedding
{
  "provider": "openai|ollama|huggingface",
  "model": "text-embedding-3-small",
  "api_key": "sk-...",
  "batch_size": 32
}

Retrieval

POST /v1/retrieve
{
  "query": "machine learning optimization",
  "top_k": 10,
  "include_graph": true
}

LLM Synthesis

POST /v1/retrieve-with-llm
{
  "query": "How does gradient descent work?",
  "top_k": 5,
  "llm_provider": "openai",
  "llm_model": "gpt-4"
}

Configuration

Environment Variables

Variable	Description	Default
DATABASE_URL	Postgres connection	postgresql://postgres:password@postgres:5432/sembra
BARQ_DB_URL	BarqDB endpoint	http://barq-db:8080
BARQ_GRAPHDB_URL	BarqGraphDB endpoint	http://barq-graphdb:8081
CELRIX_URL	Celrix cache	celrix:6380
AIMESH_URL	AiMesh queue	http://aimesh:9000
JWT_SECRET	JWT signing key	Required

Development

Backend (Rust)

cd backend
cargo build --release
cargo test

Frontend (React)

cd frontend
npm install
npm run dev

Docker Build

cd docker
docker-compose up --build -d

Project Structure

sembra/
├── backend/
│   ├── sembra-api/      # REST API server
│   ├── sembra-core/     # Business logic, AiMesh consumer
│   ├── sembra-storage/  # Postgres + BarqDB clients
│   ├── sembra-graph/    # BarqGraphDB client
│   ├── sembra-cache/    # Celrix client
│   └── sembra-types/    # Shared types
├── frontend/            # React UI
├── docker/              # Docker configuration
└── docs/                # Documentation

License

MIT License

---

Built with BarqDB, BarqGraphDB, AiMesh, and Celrix