Find your next role

Discover amazing opportunities across our network of companies committed to gender equality in the workplace.

Applied Scientist, AGI Vertical Services - Speech, Audio & Video

Amazon

Amazon

North Rhine-Westphalia, Germany
Posted on Jul 10, 2025

DESCRIPTION

The Artificial General Intelligence - Vertical Service (AGI-VS) team is seeking a dedicated, skilled, and innovative Applied Scientist with a robust background in deep learning, to build industry-leading technology with Large Language Models (LLMs) and in multi-modal systems.

Key job responsibilities
As part of the AGI-VS team, you will work on solving key industry problems through specialized and optimized models and services. You will lead the development of advanced approaches and modeling techniques, driving forward the frontier of LLM technology. This includes innovation on multi-modal (speech/audio/video) representations, understanding, transformation and generation for streaming and non-streaming applications, learning from human and AI feedback, inference time optimizations, and data efficient scaling.

A day in the life
You will support the science solution design, run experiments, research innovative algorithms, and find new ways of optimizing the customer experience, while setting examples for the team on best science practice and standards. You will work closely with talented engineers and scientists to deliver your robust solution to core customers within and outside of Amazon.
Along this you create reliable, scalable and high-performing solutions, demonstrate exceptional technical expertise, and showcase a sound understanding of Machine Learning in general and LLM technology in particular.


About the team
The AGI-VS team's mission is to solve industry- or task- specific GenAI problems through specialized models and services. Powered by Amazon AGI’s Foundational Models (e.g., Nova family of models), Compute and Modeling Infrastructure, we accelerate productivity for 1P and 3P professionals, and unbox true creativity in content creation.