Machine Learning Engineer, Amazon General Intelligence (AGI)

The Artificial General Intelligence (AGI) team is looking for a passionate, talented, and inventive Machine Learning Engineer (MLE) to play a pivotal role in the development of industry-leading multi-modal and multi-lingual Large Language Models (LLM). As our SDE/MLE superstar, you'll lead the charge in developing training algorithms and modeling techniques that will push the boundaries of large model training using GPUs and AWS Trainium. Your work will directly impact our customers' lives through game-changing products and services powered by your Generative AI breakthroughs!

Get ready to dive into Amazon's vast and diverse data sources and harness the immense power of our large-scale computing resources to turbocharge the development of multi-modal Large Language Models (LLMs) and other awe-inspiring Generative Artificial Intelligence (Gen AI) applications. Your expertise and insights will be invaluable in defining data strategies, model optimizations, and evaluation methods that will set new standards in the industry!

So, if you're passionate about pushing the limits of AI, thrive in a fast-paced and innovative environment, and are ready to make a lasting impact on the world, this is your chance! Join us on this exhilarating adventure and let's revolutionize AGI together!

• Ability to quickly learn new technologies and algorithms in the field of Generative AI to participate in our journey to build the best LLMs.
• Responsible for the development and maintenance of key platforms needed for developing, evaluating and deploying LLM for real-world applications.
• Work with other team members to investigate design approaches, prototype new technology and evaluate technical feasibility.
• Work closely with Applied scientists to process massive data, scale machine learning models while optimizing.
• Proven track record of optimizing GPU workloads at the kernel level.

A day in the life: As a SDE/MLE with the AGI team, you will be responsible for leading the development of modeling techniques and optimizing performance of the state of the art of large model training using hardware like NVIDIA GPUs. You will leverage Amazon’s heterogeneous data sources and large-scale computing resources to accelerate development with multi-modal Large Language Models (LLMs) and other Generative Artificial Intelligence (Gen AI) applications. As a key player in our team, you'll have a significant influence on our overall strategy, shaping the future direction of AGI at Amazon. You'll be driving system architecture and champion best practices that will ensure an unparalleled infrastructure of the highest quality. Work in an Agile/Scrum environment to move fast and deliver high quality software.

About the team: Join our AGI team and work at the forefront of AI. Collaborate with top minds pushing boundaries in deep learning, reinforcement learning, and more. Gain valuable experience and accelerate your career growth. This is a unique opportunity to create history and shape the future of artificial intelligence.

Mission of the team: We leverage our hyper-scalable, general-purpose large model training and inference systems to develop and deploy cutting-edge sensory AI foundational models that revolutionize machine perception, interpretation and interaction, with humans and with the physical world.

• 3+ years of non-internship professional software development experience
• 2+ years of non-internship design or architecture (design patterns, reliability and scaling) of new and existing systems experience
• Experience programming with at least one software programming language

• 3+ years of full software development life cycle, including coding standards, code reviews, source control management, build processes, testing, and operations experience
• Master's degree in computer science or equivalent
• Experience in techniques like kernel fusion and custom kernels to improve GPU utilization, mixed precision training using lower precision and dynamic loss scaling while leveraging hardware specific mixed precision capabilities and/or demonstrated ability to implement efficient memory management like gradient (activation) checkpointing, gradient accumulation, offloading optimizer states, and smart prefetching.