​Most never see their AI work make it to production.

​This invite-only evening is for AI engineers and product leads who are tired of building demos that disappear, want to work on systems that actually ship, scale, and are adopted by real users.

​In partnership with Experis, we're showcasing how we evaluate and validate two production systems that operate under the constraints of highly regulated industries. We’ll dive into how Supercell evaluates and monitors an app used by thousands of players across dozens of countries and multiple games, and how an international bank approaches evaluation and safety in novel voice-based customer-facing systems, where failure modes and constraints are very different.

​Together, they show how scale, modality, and organisational complexity reshape what “good” looks like in AI evaluation.

​We'll also lift the lid on what's driving the AI job market right now and what's actually getting engineers hired.

​If you're building AI that needs to work and deploy, not just impress in a slide deck, this evening is for you.

​Agenda:

​🍕🥤🤝 - 6pm arrival for pizza, drinks and a friendly chat

​📢 6:45pm - sessions begin

​How to Guardrail and Evaluate a Voice-to-voice System - Ale Zacarias & Andrew Finch

​There are no industry standards for evaluating voice AI in production. So we built our own.

​A practical, phased framework for testing everything from factual accuracy to whether the agent sounds human when interacting with customers. We'll cover the evaluation methods that actually worked, the ones that didn’t, and the architectural decisions we'd change if we started over.

​The AI Job Market Right Now - Experis Recruitment

​What skills get you interviews? What are companies actually paying? What are companies actually building?

​The AI job market is moving as quickly as the technology itself. While the AI Engineer might be the most in demand job right now, there are still challenges when it comes to standing out and landing the right role for you.

​Leading recruiter Experis will share what's really getting engineers hired right now, from the technical skills that differentiate candidates, the signals that get you screened out, and what to expect from a job market moving at breakneck speed.

​Scaling Agent Evaluation Across Games and Languages: How We Built Supercell’s Multi-Agent Account Recovery System - Romain Bourboulou

​We developed a virtual agent that assists players with account recovery. It drastically reduced player response time while matching, and in some cases exceeding, the performance of human specialists. It achieves this in a fraction of the time and cost of the previous solution, while keeping humans in the loop for difficult cases. In this presentation, we will:

• ​Detail the architectural choices that allow the solution to scale to thousands of tickets per day while keeping responses fast and decisions interpretable.

• ​Explain how the same architecture enables evaluation to scale across multiple games and languages and give a high-level overview of our evaluation “loop.”