Generative AI has been all the rage this past year. It can be leveraged in many creative ways that the tech community is only beginning to scratch the surface on. This talk will focus on how we at Crafted fine-tuned a Stable Diffusion model with pictures of our CEO to enable employees to prompt the model via Slack to generate fun and ridiculous photos. I will talk about the process of fine-tuning stable diffusion and focus primarily on the production operation of the model. I will discuss the many different options that one has for deploying a GPU-based inference workload to the cloud. I will go over the various options that I considered, the option that we landed on and why, and things I would do differently in the future.

Meet Your Speaker - Jeremy Gustine:
Jeremy is the Director of Engineering at Crafted, a consultancy that utilizes a balanced-team model to efficiently deliver the right product for clients. Prior to Crafted, Jeremy worked at Pivotal Labs where he originally became involved with the consulting side of software delivery. He has worked in a variety of tech stacks (including Java, .Net, Python, and Unity), and has recently started dabbling with Machine Learning and MLOps. Outside of work you can usually find Jeremy in the mountains - sliding down them in the winter and climbing them in the summer.