Machine Learning Engineer, Model Customization, Generative AI Innovation Center

Amazon•Mountain View, CA

About The Position

The Generative AI Innovation Center at AWS empowers customers to harness state of the art AI technologies for transformative business opportunities. Our multidisciplinary team of strategists, scientists, engineers, and architects collaborates with customers across industries to fine-tune and deploy customized generative AI applications at scale. Additionally, we work closely with foundational model providers to optimize AI models for Amazon Silicon, enhancing performance and efficiency. As an SDE on our team, you will drive the development of custom Large Language Models (LLMs) across languages, domains, and modalities. You will be responsible for fine-tuning state-of-the-art LLMs for diverse use cases while optimizing models for high-performance deployment on AWS’s custom AI accelerators. This role offers an opportunity to innovate at the forefront of AI, tackling end-to-end LLM training pipelines at massive scale and delivering next-generation AI solutions for top AWS clients.

Requirements

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
Bachelor's degree in computer science or equivalent
Hands-on experience with deep learning and/or machine learning methods (e.g. for training, fine tuning, and inference)
Hands-on experience with generative AI technology

Nice To Haves

3+ years of full software development life cycle, including coding standards, code reviews, source control management, build processes, testing, and operations experience
1+ years of programming with at least one software programming language experience
1+ years of experience hands-on experience with developing, deploying, or optimizing machine learning models using a recognized ML library or framework

Responsibilities

Large-Scale Training Pipelines: Design and implement distributed training pipelines for LLMs using tools such as Fully Sharded Data Parallel (FSDP) and DeepSpeed, ensuring scalability and efficiency
LLM Customization & Fine-Tuning: Adapt LLMs for new languages, domains, and vision applications through continued pre-training, fine-tuning, and Reinforcement Learning with Human Feedback (RLHF)
Model Optimization on AWS Silicon: Optimize AI models for deployment on AWS Inferentia and Trainium, leveraging the AWS Neuron SDK and developing custom kernels for enhanced performance
Customer Collaboration: Interact with enterprise customers and foundational model providers to understand their business and technical challenges, co-developing tailored generative AI solutions