Edge AI Data Scientist – Multimodal Model Evaluation

Amazon Science; Cambridge; On-site; GBP 70,000 - 120,000; Full time

Job summary

Amazon Devices is seeking an exceptional Data Scientist to join the Edge AI Product Science team. You will advance multimodal deep learning model evaluations across our devices, driving data-first improvements to model and system performance in conjunction with hardware and software teams.

You will curate datasets and develop evaluation methods for multimodal language models and agents, including audio and vision experiences, with opportunities to deploy produced insights into product use cases.

Qualifications

• Master's degree in a quantitative field such as statistics, mathematics, data science, economics, engineering, or computer science.
• Experience with machine learning/statistical modeling and data analysis tools.
• Experience applying theoretical models in an applied environment.

Responsibilities

• Collaborate with cross-functional engineers and scientists to advance multimodal model evaluations for devices (audio, images, video).
• Invent and validate automated evaluation methods for perception tasks including LLM-as-judge.
• Develop and extend evaluation frameworks for multimodal language models.
• Analyse large offline/online datasets using Python to understand model gaps and improve capabilities.
• Work with scientists, engineers, data collection and product teams to advance evaluation methods.

Skills

Multimodal ML

Data analysis

Python

Experiment design

Education

Master's degree in quantitative field

Tools

SQL

Python

R

Matlab

SAS

Job description

Amazon Devices is seeking an exceptional Data Scientist to join the Edge AI Product Science team. You will advance multimodal deep learning model evaluations across our devices, driving data-first improvements to model and system performance in conjunction with hardware and software teams.

You will curate datasets and develop evaluation methods for multimodal language models and agents, including audio and vision experiences, with opportunities to deploy produced insights into product use cases.