Diesmal: AI Model Building with LoRAs & Local LLMs mit Portrait XO *WORKSHOP in ENGLISH!* In this hands-on workshop, participants will learn how to run AI models using local large language models (LLMs) and Low-Rank Adaptation (LoRAs: a lightweight, accessible method for customizing AI behavior without the massive computational costs of training from scratch). Participants will begin by setting up a local LLM — a fully functional AI model running directly on their own machine — giving agency and privacy of local data usage. From there, participants will learn how to fine-tune that model using LoRAs: a lightweight, accessible method for customizing AI behavior without massive computational costs. We’ll walk through the full pipeline — curating ethically sourced training data with clear provenance and consent, preparing datasets, configuring LoRA parameters, running a fine-tune, and evaluating the results. The workshop will engage with deeper questions this work demands: What does it mean to own your AI infrastructure? How can small-scale model building become an act of creative sovereignty? How does running locally shift the power dynamics of who gets to shape AI — and who gets shaped by it? We’ll approach model training as a practice rooted in consent, data provenance, and intentionality — asking not just how to train a model, but what we feed it and why. No prior machine learning experience is required. Bring a laptop and a willingness to get into critical engagement with generative AI.