Modular open-source RAG platform you deploy on your own terms.
Plug in any LLM provider, vector store, or pipeline, keep data on-premises, and extend with vertical modules like e-learning, all through configuration rather than code forks.
Main monorepo. FastAPI core, document ingestion (PDF, OCR, Word, PowerPoint), vector search (pgvector or Qdrant), RAG query pipeline with optional reranker, and an e-learning vertical with embeddable chat widget. Apache 2.0.
Moodle LMS plugin. Server-side block that registers enrollments, ingests course materials, and embeds the chat widget into course pages. PHP, distributed as a standard Moodle plugin.
git clone https://github.com/vektralabs/vektra-stack.git
cd vektra-stack
cp .env.example .env
docker compose up -d
make demoFull walkthrough: getting started guide.
Unlike RAG toolkits (LangChain, LlamaIndex, Haystack), Vektra RAG ships as a deployable platform: docker compose up, configure via environment, run on your own infrastructure with built-in GDPR compliance (retention, audit logs, on-premises data).
Apache License 2.0. Maintained by VektraLabs.