Workshop- Building a RAG AI Assistant for Your Research
Generic AI chatbots can’t see your research, and when they don’t know, they often make something up. This workshop shows you how to build your own AI assistant that answers questions from your papers, reports, notes and technical documentation, and backs every answer with the sources it used.
You’ll learn Retrieval-Augmented Generation (RAG), working hands-on with open language models served through Ollama on DiRAC research-computing testbeds. By the end you’ll have a working local assistant built on your own documents.
What you’ll cover
Preparing and ingesting research documents Text extraction, chunking and embeddings Vector search and retrieval Connecting retrieved evidence to an open language model Prompt design and citation-aware responses Evaluating retrieval quality, answer faithfulness and citation accuracy Responsible use: privacy, intellectual property and research integrity Who it’s for Researchers at any career stage, from any discipline and technical background. No machine learning experience is needed.
What you’ll need
Basic Python: you can run a Jupyter notebook and follow simple code. Must: Your own laptop. A small collection of your own documents (optional; example documents will be provided). Participants will access the DiRAC testbeds over SSH. Account setup instructions and a short preparation pack will be sent after registration, and helpers will be on hand throughout the practical sessions.
Bursaries available We are offering 10 bursaries of £500 each to support attendance. To be considered, please complete the bursary section of the registration form. Bursaries will be awarded on a first-come, first-served basis, so early registration is encouraged.
Registration is required. Places are limited.
-
Format: In person
-
Location: Durham, UK
More information & registration