AI Research Engineer (Pre-training - LLM & Multi-Modal) - 100% Remote Worldwide

Tether; United States; Remote; USD 180,000 - 240,000; Full time

Job summary

Tether seeks an AI Research Engineer focused on pre-training for LLMs and multimodal models. You will design architectures, curate datasets, and optimize model performance.

The role emphasizes expertise in NLP and distributed training to scale AI across diverse modalities. You will drive innovations in model architectures for large-scale pre-training, integrating text and vision, and leveraging distributed training on thousands of NVIDIA GPUs.

Qualifications

• Strong background in NLP and distributed training.
• Experience with large-scale pre-training for LLMs.
• Experience with multimodal data and cross-modal alignment.

Responsibilities

• Design architectures and prototype models for pre-training LLMs and multimodal systems.
• Curate datasets and optimize model performance.
• Communicate findings and collaborate with a global team.
• Work with tokenizers and cross-modal alignment layers.

Skills

NLP expertise

Distributed training

Pre-training for LLMs

Multimodal modeling

Job description

Tether is seeking an AI Research Engineer to focus on pre-training for LLMs and Multi-Modal models. This role involves designing architectures, curating datasets, and optimizing model performance. Ideal candidates have a strong background in NLP and distributed training.

As an AI Research Engineer at Tether, you will drive innovations in AI model architecture specifically for large-scale pre-training of LLMs and Multi-Modal systems. Your work will involve integrating diverse modalities like text and vision, utilizing advanced distributed training on thousands of NVIDIA GPUs.

You will be responsible for designing and prototyping architectures, tokenizers, and cross-modal alignment layers to enhance model capabilities. Success in this role requires not only technical expertise in AI and ML but also the ability to effectively communicate findings and collaborate with a global team.