Most “AI agent” roles are prompt engineering with better branding. This isn’t one of them. We’re building agents that actually operate inside live games: - Reason over images and video across frames - Decide what to do next in messy, shifting UIs - Tap, swipe, navigate, recover from failure - Run reliably across devices, lag, edge cases No fixed scripts. No demo theatre. Production systems that studios trust more than manual QA. You’ll take our proprietary VLM and turn it into a real-world decision-making system. Think less “clever prompts” and more autonomy, control loops, failure handling, systems design. If you’ve built: - Vision-based agents - Robotics / autonomy systems - Complex UI automation that had to survive the wild - Production ML systems where reliability actually mattered We should talk. Founding-level ownership. Early traction. Real customers. *Please note that this role requires you to be based in London, UK.