Machine Learning Engineer @ LoadComplete

Full-time; 2022/07 - Present

Currently working at LoadComplete’s Applied ML team as a machine learning engineer, focusing on building and integrating machine-learning-based solutions into the studio’s game development processes.

• Led the design and development of the machine learning-based automated playtesting process for Frame Arms Girl: Dream Stadium, which greatly contributed to bug discovery and design validation. This involved devising and implementing all of the system components behind this process and the model’s architecture.
• Designed and implemented the internal solution for building and deploying deep learning-trained game-playing agents in Unity. This solution addressed the limitations of the previous solution by featuring improved speed, scalability, and extensibility through features such as async action retrieval, dynamic batching, a plugin-based architecture, and more robust fault tolerance.
• Implemented advanced training algorithms and techniques for optimized and more effective model training.
• Currently developing ML-based automated playtesting for an upcoming title, and actively experimenting with ideas related to player modeling and game personalization to improve user experience and create business value.

GDC 2024

I presented the work on the automated playtesting process for Frame Arms Girl: Dream Stadium at GDC 2024 in a talk titled “Testing Empowered: Integrating ML-Based Playtesting in a Team with Limited QA Capacity”. Recording of the talk can be found here.

Skills

PyTorch, Python, Unity, C#, Git, AWS / GCP, Deep Reinforcement Learning