Evals London #1
Samuel Colvin opens the first Evals London.
Samuel created Pydantic and now runs the company behind Pydantic AI and Logfire. His keynote is about evaluation and prompt optimization.
Running order
18:00 Check-in, food and drinks
18:30 Welcome, then Samuel's keynote
19:15 Discussion tables
20:15 (about) Networking
21:00 Close
Discussion tables
1. LLM evals 101, with Šimon Podhajský
2. Building evals with domain experts, with Conor Svensson
3. Agent and eval debugging, with Reza Amini
4. Evaluating open-ended research agents, with Giovanni Riva
5. Long-running agent swarm evals, with Lindell Cumes
6. Trusting evals, with Jenny Chim
Pick a table after the keynote. Each table lead opens with a few minutes on how they approach the problem, then the group works through its own cases. No slides. Bring the eval problem you haven't cracked yet, say a judge you only half trust, or a test set that quietly stopped catching regressions.
Who should come: engineers and data scientists building on LLMs, product people who own AI quality, and researchers. No sales pitches, from the stage or in the groups.
Evals London is the sibling of evals.cz, which has run four meetups in Prague since February. All filled up and ran waitlists. Places here are limited too, so if you register and your plans change, please cancel so someone on the waitlist gets your spot.
Hosted by Dawn Capital, Europe's leading specialist B2B software investor, at their Soho office. Catering kindly sponsored by Pydantic.
Check-in opens at 18:00 in the lobby of Ilona Rose House. Building security needs your full name in advance, so please register with your first and last name. By registering you agree to Dawn Capital's event terms: https://drive.google.com/file/d/1u1i2WKT0NeoV-YslA-tcN4-xVO5CYhXv/view