I'm the Tech Lead of the Discovery Team at Google DeepMind, where I lead post-training research for large language models focused on creativity, reasoning, and scientific discovery.

My research interests center on reasoning agents, world models, combinatorial optimization, and interpretability. I'm particularly focused on building AI systems that can generate novel solutions and push the boundaries of machine creativity. I also think deeply about AI alignment and long-term AI strategy.

I share technical insights and research reflections on my blog and maintain an incomplete list of open-source projects I've built.

Previously, I worked on PCB routing using reinforcement learning at InstaDeep (acquired by BioNTech). I studied Electrical and Computer Engineering at the University of Cape Town, with an emphasis on control systems and signal processing.

Contact me at ttumiel[at-gmail-dot-com] or DM me on twitter.