Senior Multimodal AI Engineer

Cohere; New York (NY); On-site; USD 150,000 - 210,000; Full time

Benefits offered by this job

Health benefits

Remote-friendly culture

Learning stipend

Job summary

Cohere, a remote-friendly AI security-first company, seeks a member of technical staff focused on multimodal AI. Design and build systems that combine text, speech, and vision while exploring new representation learning ideas and transfer learning.

You will work with top teams to push the boundaries of model performance and real-world applicability. Ideal candidates have strong Python skills, experience with JAX/PyTorch/TensorFlow, and a track record in distributed training for large-scale

Qualifications

• Proven software engineering skill with a track record of building scalable systems.
• Strong Python proficiency and experience with DL frameworks (JAX, PyTorch, TensorFlow).
• Expertise in distributed training for large multimodal models.
• Familiarity with autoregressive approaches for multimodal tasks and GPU kernel development is a plus.

Responsibilities

• Design and develop cutting-edge multimodal AI systems integrating text, speech, and vision.
• Conduct research and experiments on advanced compute infrastructure and representations.
• Collaborate with world-class teams to advance Cohere's frontier models.

Skills

Python

JAX

PyTorch

TensorFlow

Distributed training

Multimodal models

CUDA

Tools

CUDA

Job description

Cohere, a remote-friendly AI security-first company, seeks a member of technical staff focused on multimodal AI. Design and build systems that combine text, speech, and vision while exploring new representation learning ideas and transfer learning.

You will work with top teams to push the boundaries of model performance and real-world applicability. Ideal candidates have strong Python skills, experience with JAX/PyTorch/TensorFlow, and a track record in distributed training for large-scale