Remote AI Research Engineer – Model Compression & Edge AI

Tether Ltd.; Greater London; Hybrid; GBP 90,000 - 150,000; Full time

Job summary

Tether Ltd. is seeking an AI researcher to advance model compression for multimodal systems, including large language models and vision-language models.

You will reduce footprint and computational cost while preserving accuracy, applying quantization, distillation, and pruning to complex architectures that integrate text, images, and audio. You will build robust compression pipelines, establish performance metrics, and drive production-ready inference on edge devices.

Qualifications

• Experience with state‑of‑the‑art NLP and multimodal models.
• Proven research and development track record in AI R&D (publications a plus).
• Strong publication/presentation skills in top-tier venues.

Responsibilities

• Apply low-bit quantization to reduce model size and inference latency for multimodal AI models while maintaining accuracy.
• Leverage knowledge distillation to transfer capabilities from larger teacher models to smaller student models.
• Implement pruning techniques to remove redundant parameters and attention heads, reducing computational overhead.
• Analyze trade-offs between model efficiency and accuracy across quantization, distillation, and pruning methods; propose improvements based on empirical findings.
• Research and apply mixed‑precision quantization and other advanced compression strategies to optimize accuracy–performance balance.
• Stay current with the latest research in model compression for multimodal and generative architectures.
• Document methodologies, experiments, and results to support reproducibility and collaboration.
• Author technical papers and publish findings in top-tier conferences to advance the field.

Skills

PyTorch

Quantization

Distillation

Pruning

Transformers

C++

Education

PhD in NLP/ML

BSc in Computer Science or related

Tools

PyTorch

C++

Job description

Tether Ltd. is seeking an AI researcher to advance model compression for multimodal systems, including large language models and vision-language models.

You will reduce footprint and computational cost while preserving accuracy, applying quantization, distillation, and pruning to complex architectures that integrate text, images, and audio. You will build robust compression pipelines, establish performance metrics, and drive production-ready inference on edge devices.