Dinner: People behind AI |London

​Benchmarks get the headlines. The people who build them rarely get a seat at the table.

​People behind AI is a dinner series by Sovrano for the researchers doing the work underneath frontier models: evaluation, post-training, RL environments, data. The work that decides whether a model is actually good, and that almost never makes the launch post.

​After Madrid, San Francisco and Lisbon, we’re bringing it to London.

​Twelve seats. One table. No talks, no panels, no slides.

​Just a good dinner and the conversations you can’t have on X or at a poster session:

• ​What are we actually measuring when we say a model got better?

• ​Where do evals break, and what replaces them?

• ​What can expert human judgment do that synthetic data still can’t?

• ​Which problem in your field is everyone quietly avoiding?

​Who it’s for
Researchers and engineers working on evaluation, post-training, RL, alignment or data, at frontier labs, AI startups and universities. If you’ve ever argued about a benchmark at 1am, you’ll fit right in.

​Details
Thursday 12 November, 19:00
Restaurant at Soho. Venue shared with confirmed guests.
Dinner is on us ;)

​Seats are limited and approved by request. Tell us briefly what you’re working on.

​Hosted by Sovrano. We build expert data infrastructure for frontier AI: the evaluations, environments and human expertise that models are measured against.