Advancing HPC skills by Developing Large Language Model Retrieval Augmented Generation (LLM-RAG) Systems

Volume 17, Issue 1 (March 2026), pp. 28–33

Download PDF

@article{jocse-17-1-4,
  author={Julia Mullen and Sam Corey and Lauren Milechin and Riya Tyagi and Daniel Burrill},
  title={Advancing HPC skills by Developing Large Language Model Retrieval Augmented Generation (LLM-RAG) Systems},
  journal={The Journal of Computational Science Education},
  year=2026,
  month=mar,
  volume=17,
  issue=1,
  pages={28--33},
  doi={https://doi.org/10.22369/issn.2153-4136/17/1/4}
}

Large Artificial Intelligence (AI) and generative large language models (LLM) are key computational drivers. For researchers developing new tools or incorporating LLMs into their processing pipeline, the scale of data and models require supercomputing resources which can only be met through cloud or High Performance Computing (HPC) architectures. Many of these researchers have deep experience with AI, LLMs, and their research area but are new to HPC concepts, challenges, tools, and practices. To assist this researcher community, the Research Facilitation Teams at MIT Office of Research Computing and Data (ORCD) and the MIT Lincoln Laboratory Supercomputing Center (LLSC) have developed tutorial materials to teach researchers how to build their own Retrieval Augmented Generation (RAG) workflows. Selecting RAG systems as the project focus provides motivation for developing a wide range of skills necessary for efficiently working with LLMs on an HPC system while creating a useful application. This work details LLM-RAG implementation concerns on two different systems, the design decisions associated with developing the examples, deployment of the workshop training, and the feedback received from the participants. Both the MIT ORCD and MIT LLSC systems are representative of HPC community systems and we plan to refactor the in-person and live virtual workshops into a micro-course built from online, self-paced modules that will be reusable across other HPC centers with slight modifications.