View, browse and sort the ever-growing list of sessions by pass type, track, and format. With this Session Viewer, you can view session and speaker details for the 2024 Game Developers Conference. New sessions are regularly added leading up to GDC, and all dates and times will be announced about 4 weeks before the event. Once live, you will be able to build your schedule with the GDC Mobile App.
Sessions do fill up and seating is first come, first serve, so arrive early to sessions that you would like to attend.
Sun Jeong (Machine Learning Engineer, LoadComplete)
Pass Type: All Access Pass, Core Pass - Get your pass now!
Vault Recording: TBD
Audience Level: Intermediate
Playtesting is a crucial aspect of modern game development. However, small-to-mid-sized teams often face difficulties in meeting their testing demands due to limited QA resources.
In this talk, Sun Jeong discusses how he and his team built and harnessed a machine-learning-assisted automated playtesting solution to fulfill the testing requirements for the launch and live service of an action-packed roguelite game, Frame Arms Girl: Dream Stadium.
He talks about the technical details of the automated playtesting stack, introduces various detailed use cases and the workflow used to debug and discover valuable design insights, and shares the tips and lessons learned during the development of the stack under dynamic and unstable production settings.
Attendees will gain inspiration and develop a better understanding of how teams of mid-to-small sizes can build and benefit from machine-learning-powered automated playtesting by learning from various detailed use cases and technical breakdowns. They will also learn tips on preparing agents under production conditions.
This is for developers who are interested in deploying a similar machine learning-powered playtesting solution and want to learn detailed examples of how these solutions can be used in real-world settings for debugging, design insight discovery, and more. Familiarity with deep learning and reinforcement learning is strongly advised.