Senior Applied Scientist

What you'll do

• Design, implement, and evaluate mid-training approaches to improve multimodal generative model editing capabilities.
• Own specific components of the mid-training stack, such as image-to-image or instruction-based editing.
• Design and run experiments to test hypotheses and identify quality gaps in generative models.
• Build and maintain large-scale captioning pipelines to improve multimodal understanding across visual and auditory domains.
• Support VLM finetuning efforts to enhance model performance for downstream multimodal tasks.
• Build scalable workflows for data curation, quality improvements, and distributed training.
• Apply research insights from diffusion models to practical large-scale model improvements.
• Document experiments, datasets, and training approaches for cross-functional knowledge sharing.

What we're looking for

• Possession of a Master’s or Ph.D. degree in Computer Science, Machine Learning, or a related field.
• Solid understanding of modern generative architectures such as diffusion models with familiarity with conditional generation or editing methods.
• Experience implementing machine learning models using modern deep learning frameworks like PyTorch.
• Experience with large-scale or distributed training workflows.
• Research background in VLM finetuning with a focus on image, video, and audio understanding.
• Familiarity with designing or applying automated captioning pipelines for large-scale image, video, and audio datasets.
• Strong coding and prototyping ability in Python and PyTorch.
• Excellent communication skills and ability to collaborate across cross-functional teams.