Preparing Data for LLMs

On Thursday, November 19th, Nicolo’ Brandizzi will be presenting “From 173 Billion Documents to Training Data: A Beginner’s Tour of LLM Data Processing” at Conf42’S MLOPS 2026. Join us to learn about Nicolo’s work at Frauenhofer IAIS, and workshop material based off of this research that is now available as open course content. Learn step-by-step how to take data from the raw Common Crawl data dump and proceed through all of the steps needed to make it ready to send off to an LLM for training. By the end you will know how a training corpus is built, and what that tells you about the models trained on it. No data-pipeline experience needed.