Member of Technical Staff Pretraining

You will lead large-scale pretraining experiments for image, video, and audio foundation models. You will develop and evaluate architecture, optimizer, and training-algorithm ideas; contribute GPU and systems optimizations and research code; and independently lead research projects and cross-team initiatives.

Responsibilities

• Lead large-scale multimodal foundation-model pretraining experiments
• Develop and evaluate architectures, optimizers, and training algorithms
• Contribute GPU and systems optimizations, research code, and model design
• Lead independent research projects and cross-team initiatives

Requirements

• Experience leading or co-owning foundation-model pretraining through production or a major release
• Experience with architecture, objective, attention, modulation, loss, and tokenization design
• Experience with large-scale distributed training on multi-node GPU clusters
• Python and PyTorch proficiency
• Familiarity with visual generative models
• Production impact or top-venue research publications

Benefits

• Equity
• Reasonable travel costs covered