Research Engineer - Multimodal AI

Responsibilities

• Develop, implement, and maintain MoE models, data pipelines, batch inference operating at internet scale, processing large, multimodal datasets including images, text, and videos.
• Integrate the latest research methods into our multimodal models and data distillation platform, ensuring high standards of data accuracy, relevance, and diversity.
• Innovate and experiment with new SOTA models and data curation techniques to maximize the quality and efficiency of training advanced multimodal models.
• Continuously improve data & AI infrastructure to support scalability and flexibility as models and data requirements evolve.

Qualifications

• Bachelor’s or advanced degree in Computer Science or a related field.
• Proficiency in Python and experience with large open source datasets like DataComp.
• Solid understanding of distributed computing and experience working with large-scale, high-throughput systems.
• Hands-on experience with visual data and multimodal model training.
• Familiarity with deep learning frameworks, especially for handling multimodal data in model training.
• Strong interest in large-scale visual model research and comfort working in a rapidly evolving, dynamic environment.