Responsibilities:
Design industry-first evaluation suites and reward models using Reinforcement Learning from Human Feedback (RLHF) and gameplay telemetry to systematically quantify subjective game quality, balance, and core loop "fun," while deploying autonomous multi-agent frameworks to playtest generated code and verify physical logic before release.
Address the spatial and interactive blind spots of standard Large Language Model (LLMs) by fine-tuning models with deep domain knowledge in three-dimension (3D) mathematics, engine architectures (e.g., Unity, Godot, Unreal), and physical reasoning (e.g., gravity, momentum, collisions), building frictionless bridges directly into real-time render loops.
Build high-efficiency two-dimension (2D)/3D asset pipelines using fast distillation and DiT models to instantly generate sprites and rigged 3D meshes.
Build multi-agent Artificial Intelligence (AI) frameworks to automatically playtest generated levels for physics errors, boundary bugs, and overall playability.
Minimum qualifications:
Bachelor’s degree or equivalent practical experience.
5 years of experience in Machine Learning or Artificial Intelligence.
Experience with coding in data structures, algorithms and software design.
Preferred qualifications:
8 years of experience in software development
Ready to apply?
Sign up first — takes a minute — and you get Hiro’s take on this role, a resume tailored to it, and (if available) a referral from a real employee at Google. All free with your Pro gift.
Sign up to apply