AI Systems, Training

About the role

Since 2022, AI has become mainstream, transforming industries and creating a need for more efficient computation. At Unconventional, we aim to address this by building a new foundation for AI that is 1000x more efficient. Our mission is to rethink computing from the ground up, exploiting the physics of semiconductors to map neural networks directly to device physics.

Responsibilities

• Build and maintain highly optimized, model-specific training stacks specifically tuned for state-of-the-art generative vision, language, and world models.
• Design and scale multi-node distributed training systems, implementing elastic sharding and robust data streaming pipelines for fast, large-scale iteration.
• Implement and robust model checkpointing and recovery mechanisms.
• Develop and optimize kernels using low-level programming models like CUDA and Triton.
• Design rigorous benchmarking suites to track Model Flops Utilization (MFU), memory bandwidth, and convergence stability.
• Act as a translator, discussing algorithmic trade-offs with theorists and converting model requirements into concrete specifications for infrastructure and hardware engineering teams.

Requirements

• Education: An MS/PhD or equivalent research/project experience in a quantitative field such as AI/Machine Learning, Computer Science, Physics, Electrical Engineering, or Applied Math.
• Experience: Veteran of the modern ML software stack. Demonstrated ability to map state-of-the-art AI model architectures (e.g., transformers, Mixture of Experts, diffusion models) to system performance implications. Deep expertise in how models are partitioned across a cluster, with a mastery of communication primitives, and parallelism strategies.
• Software Development: Proven track record of implementing, debugging, and maintaining production-grade training frameworks—such as Megatron-LM, DeepSpeed, Ray, PyTorch Lightning—turning raw compute into a reliable model-building factory.

Qualifications

• Education: An MS/PhD or equivalent research/project experience in a quantitative field such as AI/Machine Learning, Computer Science, Physics, Electrical Engineering, or Applied Math.
• Experience: Veteran of the modern ML software stack. Demonstrated ability to map state-of-the-art AI model architectures (e.g., transformers, Mixture of Experts, diffusion models) to system performance implications. Deep expertise in how models are partitioned across a cluster, with a mastery of communication primitives, and parallelism strategies.
• Software Development: Proven track record of implementing, debugging, and maintaining production-grade training frameworks—such as Megatron-LM, DeepSpeed, Ray, PyTorch Lightning—turning raw compute into a reliable model-building factory.

Skills

• Model Architectures: Build and maintain highly optimized, model-specific training stacks specifically tuned for state-of-the-art generative vision, language, and world models.
• Training Infrastructure: Design and scale multi-node distributed training systems, implementing elastic sharding and robust data streaming pipelines for fast, large-scale iteration.
• Optimization & Benchmarking: Develop and optimize kernels using low-level programming models like CUDA and Triton. Design rigorous benchmarking suites to track Model Flops Utilization (MFU), memory bandwidth, and convergence stability.
• Cross-Functional Collaboration: Act as a translator, discussing algorithmic trade-offs with theorists and converting model requirements into concrete specifications for infrastructure and hardware engineering teams.

Benefits

A comprehensive package including best-in-class health benefits, 401k matching, truly unlimited PTO, and complimentary meals in our Palo Alto office.

Pay

Details TBD

Schedule

Details TBD