I'm thinking of moving into a different research direction in the future. Of course, my thesis is still going to be on Gaussian processes and I still have a couple of projects to complete. But overall, I'm thinking about how to lay the groundwork for my next step. I'm hesitating between several directions: LLM (text to image/video), Bayesian optimisation, generative methods (diffusion and flow matching), ML4Science (drug discovery, climate science, etc.) and ML4Code. Any thoughts?