Remote AI Research Engineer: LLM & Multimodal Pre-Training

Tether Ltd.; Dubai; Remote; AED 600,000 - 1,000,000; Full time

Job summary

Tether Ltd. is seeking a senior AI research engineer to drive large-scale pre-training for LLMs and multimodal models within a distributed GPU environment. You will contribute to architecture, data pipelines, and optimization across a global, remote team.

The role emphasizes cutting-edge transformer and non-transformer innovations, PyTorch and Hugging Face proficiency, and a research-driven approach to push cross-modal AI capabilities.

Qualifications

• PhD in NLP or ML with AI R&D publications preferred.
• Hands-on experience with large-scale LLM or multimodal pre-training.
• Familiarity with distributed training frameworks and tooling.
• Deep knowledge of transformer and non-transformer models for efficiency.
• Strong skills in PyTorch and Hugging Face libraries.

Responsibilities

• Large-Scale Pre-Training: Conduct foundational pre-training for LLMs and Multi-Modal models on large, distributed servers with thousands of NVIDIA GPUs.
• Architecture & Alignment Innovation: Design, prototype, and scale innovative architectures and cross-modal alignment layers.
• Data Strategy: Source, filter, and curate massive-scale textual and multi-modal datasets with robust data pipelines.
• Experimental Research: Execute experiments, analyze results, and refine training methodologies for performance.
• Optimization & Debugging: Identify and fix bottlenecks in model efficiency and multi-modal alignment during long runs.
• System Scalability: Contribute to distributed training systems for scalable, hardware-efficient training.

Skills

LLM & multimodal architectures

Pre-training optimization

Distributed training

Data curation & pipelines

PyTorch

Hugging Face

Education

PhD in NLP / ML (preferred)

Tools

NVIDIA GPUs

Distributed training frameworks

Job description

Tether Ltd. is seeking a senior AI research engineer to drive large-scale pre-training for LLMs and multimodal models within a distributed GPU environment. You will contribute to architecture, data pipelines, and optimization across a global, remote team.

The role emphasizes cutting-edge transformer and non-transformer innovations, PyTorch and Hugging Face proficiency, and a research-driven approach to push cross-modal AI capabilities.