VikingRAG
Copyright 2026 Muhammad Huzaifa and contributors

This product is independently engineered software inspired by the VikingRAG
research paper. It is not a copy, fork, or redistribution of the official
VikingRAG / OpenViking research codebases (which are separately licensed).

------------------------------------------------------------------------------
Academic attribution (required when citing this work or its methods)
------------------------------------------------------------------------------

This implementation is inspired by concepts described in:

  Peiyuan Gao et al.,
  "VikingRAG: Accurate and Token-efficient Retrieval-augmented Generation
   over Structured Documents",
  arXiv:2609.11390, 2026.
  https://arxiv.org/abs/2609.11390

If you use VikingRAG in academic work, please cite the paper above
(see also CITATION.cff). Citing this repository in addition to the paper
is appreciated but optional.

------------------------------------------------------------------------------
Third-party notices
------------------------------------------------------------------------------

This project depends on open-source libraries distributed under their own
licenses (e.g. FastAPI, SQLAlchemy, Pydantic, pgvector, Redis client, pypdf,
tiktoken). See those packages for their license terms.
