Member of Technical Staff VLM

You will lead development and training of multimodal vision-language models. You will design fine-tuning strategies for creative use cases, research VLM and LLM integrations with diffusion and flow pipelines, and evaluate emerging multimodal architectures to turn research into practical improvements.

Responsibilities

• Lead development and training of multimodal vision-language models
• Design fine-tuning strategies for specialized creative use cases
• Research VLM and LLM integrations with diffusion and flow pipelines
• Evaluate emerging multimodal architectures and implement practical improvements

Requirements

• Experience pretraining or significantly advancing a deployed or publicly released vision-language model
• Record of advancing multimodal architectures through publications or production work
• Knowledge of tokenization, alignment, grounding, cross-modal attention, and failure modes
• Experience with multi-node distributed training
• Experience with diffusion or flow-based generative models is preferred

Benefits

• Equity
• Reasonable travel costs covered