AI Research Engineer (Model Compression & Quantization)

Tether Operations Limited

4d • Remote

About The Position

As a member of Tether’s AI research team, the AI Research Engineer will drive innovation in model compression and efficient deployment for advanced multimodal AI systems, including large language models (LLMs) and vision-language models (VLMs). The work focuses on reducing model footprint and computational cost while preserving accuracy, enabling high-performance AI to run efficiently on resource-constrained edge devices such as smartphones.

The role applies and advances quantization, knowledge distillation, and pruning to streamline multimodal architectures integrating text, images, and audio. It calls for a hands-on, research-driven approach to developing, testing, and implementing compression strategies that balance model size, latency, throughput, and accuracy. The engineer will build compression pipelines, establish performance and fidelity metrics, and address production inference bottlenecks, with the goal of delivering scalable, low-memory, low-latency AI systems that maintain high fidelity and real-world value.

Requirements

• Degree in Computer Science or a related field.
• Experience with PyTorch or equivalent deep learning frameworks.
• Hands-on experience with model quantization, including Quantization-Aware Training (QAT) and Post-Training Quantization (PTQ).
• Research and hands-on experience with knowledge distillation for compressing large models into smaller, efficient models.
• Research and hands-on experience with model pruning for compressing large models into smaller, efficient models.
• Solid understanding of neural network architectures and training processes, including transformers such as LLMs and VLMs, backpropagation, optimization, and fine-tuning techniques.

Nice To Haves

• PhD in NLP, Machine Learning, or a related field.
• Solid track record in AI R&D, including good publications in A conferences.
• Familiarity with C++, especially for implementing low-level quantization kernels or inference optimizations.

Responsibilities

• Apply low-bit quantization to generative AI models, including LLMs, VLMs, and multimodal models, to reduce model size and inference latency while maintaining accuracy and output quality.
• Use knowledge distillation to transfer capabilities from larger teacher models to smaller student models for efficient multimodal reasoning across text, image, and audio inputs.
• Implement pruning techniques to remove redundant parameters and attention heads, reducing computational overhead without sacrificing task performance.
• Analyze trade-offs between model efficiency, including size, latency, and memory, 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, such as adaptive pruning schedules and distillation with intermediate feature matching.
• Stay current with research in model compression, including emerging techniques for multimodal and generative architectures.
• Document methodologies, experiments, and results to support reproducibility, internal collaboration, and stakeholder communication.
• Author technical papers and publish findings in top-tier conferences such as NeurIPS, ICML, ICLR, CVPR, ACL, and AAAI.

What This Job Offers

• Job Type: Full-time
• Career Level: Mid Level