The Role
As a Member of Technical Staff, you will lead the implementation and scaling of core model architectures. You sit at the intersection of mathematical research and high-performance engineering, responsible for turning theoretical world-model hypotheses into robust, industrial-grade reality.
You own the bridge between "it works on paper" and "it works at scale." You have a deep understanding of the deep learning stack and are capable of contributing to the architectural design, implementation, and verification of models that predict the evolution of the physical world.
Job Responsibilities
- Area: Implementation and evolution of core model architectures, ensuring algorithmic designs are optimized for large-scale execution.
- Ownership: The technical performance and predictive fidelity of the World Model throughout the training and evaluation lifecycle.
- Focus: Developing high-performance implementations of complex architectures (e.g., Diffusion, Transformers, SSMs) and optimizing the training dynamics required for long-horizon rollouts.
- Operations: Establishing "Scientific Engineering" standards—creating the diagnostic tools and instrumentation that ensure every training run provides maximum insight and reproducible results.
Minimum Qualifications
- Extensive experience in software engineering for deep learning R&D teams.
- Expert-level experience implementing and optimizing large-scale generative model architectures.
- Expert with Python and C++ / Rust.
Preferred Qualifications
- Proof of outstanding contribution to open-source libraries for model implementation or research frameworks.
- Peer-reviewed research contribution in deep learning (from a AAA conference) with a focus on model architecture or training efficiency.
- Experience with hardware-aware implementation (e.g., Triton or low-level kernel optimization).