Model Engineer, Imagine

Sourced from genai.jobs

Join xAI's Imagine Model Team as a multimodal model engineer building high-fidelity image, video, and audio understanding and generation. You will own the full stack from data curation and pre/post-training to inference serving and product integration. Work closely with product teams to push model frontiers and deliver end-to-end user experiences.

• Drive engineering agendas for multimodal image/video generation, editing, understanding, and world simulation
• Improve data quality via annotation, filtering, augmentation, synthetic generation, and captioning
• Design evaluation frameworks, benchmarks, evals, and reward models for visual/audio quality
• Implement efficient algorithms for real-time inference, distillation, and scalable serving
• Build scalable multimodal data collection and processing pipelines
• Integrate AI solutions into production and iterate on user feedback

• Production-quality coding in Python, PyTorch, JAX, Rust, or C++
• Experience with large-scale distributed machine learning systems
• Data-driven experiment design and iterative model debugging
• Hands-on contributor with strong communication and prioritization skills

• Experience with SFT, RL, evals, human/synthetic data collection, or agentic systems
• Proficiency in JAX/XLA, Spark, Ray
• Domain expertise in graphics engines, rendering, image/video generation, world models, or real-time simulation
• Audio/speech processing or music generation experience
• Agentic RL training or multimodal agents