Edge AI Scientist: Compress & Deploy Multimodal Models

Eworker; United Kingdom; On-site; GBP 90,000 - 130,000; Full time

Job summary

Amazon Devices is seeking an Applied Scientist to join Edge AI Model Studio. You will apply and extend compression recipes for edge and cloud deployment across multimodal models, and design healing strategies to recover accuracy after compression.

You will collaborate with platform and hardware teams, and publish notable results at top ML conferences when possible. This role offers a fast-paced, edge-focused research environment in the United Kingdom.

Qualifications

• Master’s degree, or a PhD in CS, CE, ML or related field.
• Experience programming in Java, C++, Python or related language.
• Publications or patents in top-tier conferences/journals.
• Experience with state-of-the-art DL model design, training, optimization, and model pruning.
• Preferred experience with multimodal models, compression techniques, mixed-precision training, edge deployment, and large-scale ML systems.

Responsibilities

• Apply and extend compression recipes to assigned models to achieve large compression while preserving quality.
• Design and run healing recipes to recover accuracy after compression.
• Track model architectures and decide where to apply compression and how to recover.
• Build a library of model entries with recipes, model cards, and benchmarks for self-service deployment.
• Define datasets, benchmarks, and KPIs to evaluate accuracy, latency, memory, and cost.
• Collaborate with scientists, engineers, and partner teams; mentor interns.

Skills

Python

C++

Java

Deep learning

Multimodal models

Education

Master’s degree or PhD in CS/CE/ML

Tools

TensorFlow

PyTorch

Quantization tooling

Job description

Amazon Devices is seeking an Applied Scientist to join Edge AI Model Studio. You will apply and extend compression recipes for edge and cloud deployment across multimodal models, and design healing strategies to recover accuracy after compression.

You will collaborate with platform and hardware teams, and publish notable results at top ML conferences when possible. This role offers a fast-paced, edge-focused research environment in the United Kingdom.