Jobs at Roam (roam.fr)

Engineering Intern will build AI-powered game generation pipelines, optimize model accuracy and inference speed across mobile and cloud systems, and develop feedback loops using player and self-play data. The role involves agent orchestration, LLM post-training, evaluation harnesses, real-time game improvement, multimodal data pipelines, and context-aware creation experiences. Candidates may own major parts of the AI generation and simulation stack while helping define benchmarks for engagement and interactive intelligence in 3D worlds.

Roam is seeking exceptional talent to build AI systems that enable anyone to create games and interactive 3D worlds without technical skills. The role is intentionally open-ended, targeting candidates who can demonstrate outstanding abilities and justify a role on the applied AI team. Full-time work and residence in San Francisco, California are required.

Build applied AI systems for 3D world generation, including distributed multi-agent orchestration, reinforcement learning pipelines, multimodal data systems, diffusion-based tooling, and inference optimization. Develop scalable world-generation and simulation infrastructure, define benchmarks for interactive intelligence and engagement, and ship production systems for AI-generated games and environments.

Build and own scalable infrastructure, self-service developer platforms, AI-generation pipelines, cloud integrations, testing systems, and production toolchains. The role spans AWS infrastructure, applied AI, vector search, 3D engines, mobile platforms, and deployment automation. Responsibilities include improving architecture, defining service contracts, reducing deployment friction, and delivering reliable platform features end to end. Preferred experience includes diffusion and language model training, robotics, research, and game development.