Remote AI Research Engineer: Model Compression & Multimodal

Tether Ltd.; Abu Dhabi; Remote; AED 400,000 - 650,000; Full time

Job summary

Tether Ltd. is seeking an AI Research Engineer focused on model compression for multimodal AI systems, including LLMs and VLMs.

You will design and implement techniques to reduce model size and computation while preserving accuracy, enabling edge deployment across devices. The role requires hands-on expertise in quantization, knowledge distillation, and pruning, with a strong background in neural networks and transformers.

Qualifications

• Experience with PyTorch deep learning frameworks or equivalent frameworks.
• Hands-on experience with model quantization including QAT and PTQ.
• Research and hands-on experience with knowledge distillation for compressing large models.
• Research and hands-on experience with model pruning for compressing large models.
• Solid understanding of neural network architectures and training processes, including transformers.
• Familiarity with C++ is a plus for implementing low-level quantization kernels.

Responsibilities

• Apply low-bit quantization to reduce model size and inference latency for generative AI models.
• Leverage knowledge distillation to transfer capabilities from larger to smaller models.
• Implement pruning techniques to remove redundant parameters and attention heads.
• Analyze trade-offs between efficiency and accuracy across methods; propose improvements.
• Research and apply mixed-precision quantization and advanced compression strategies.
• Stay current with latest research in model compression for multimodal AI.
• Document methodologies, experiments, and results clearly for reproducibility.
• Publish findings in top-tier conferences to advance multimodal AI compression.

Skills

PyTorch

Quantization

Knowledge distillation

Model pruning

Transformers

C++ familiarity

Education

PhD in NLP / ML or related

Bachelor in Computer Science

Tools

Python

LLMs

Job description

Tether Ltd. is seeking an AI Research Engineer focused on model compression for multimodal AI systems, including LLMs and VLMs.

You will design and implement techniques to reduce model size and computation while preserving accuracy, enabling edge deployment across devices. The role requires hands-on expertise in quantization, knowledge distillation, and pruning, with a strong background in neural networks and transformers.