Hey everyone! In my keynote for this hackathon, I dived into how we can find new materials twice as fast and at a much lower cost using Bayesian optimization. This idea isn't new—it's actually inspired by the Materials Genome Initiative. My talk revolved around how computational and data-driven approaches in material science can drastically speed up our discovery process. I shared an example from our own research where we were on the lookout for super-hard materials. Using DFT data as a proxy for hardness, we managed to find two superhard materials in just six months, which was pretty incredible. I stressed the importance of having good data to make these data-driven methods work. Without it, we're kind of stuck. Then, I broke down the nuts and bolts of Bayesian optimization for you guys, explaining how we use it to navigate the complex landscape of material properties. We touched on surrogate models, the challenges of dealing with uncertainty, and the balance between exploring new possibilities and exploiting known ones. I also talked about the tricky business of optimizing for more than one goal at a time, like when we want a material that's both strong and cheap. It's not straightforward, but with tools like Pareto fronts, we can find the best trade-offs. Wrapping up, I encouraged everyone here to dive into the deep end with these methods during the hackathon. There are so many interesting problems to tackle, and with over 400 of you signed up, I'm excited to see what we can achieve. Let's push the limits of what Bayesian optimization can do in material science. Can't wait to see your projects and answer your questions!