Applied Scientist II, AWS Neuron Science - Core Algorithm
Core
Design and optimize hardware-aware machine learning algorithms for AWS Trainium and Inferentia accelerators, bridging LLM modeling, large-scale training systems, and custom silicon co-design.
Role type
Senior Applied Scientist (Hardware-Aware ML & Custom Silicon)
Builds
Production-ready low-precision training/inference recipes, Trn-friendly model architectures, and system-aware optimizers for AWS ML chips.
Domain
Cloud Computing / Machine Learning / Custom Silicon Co-design (via careerplan.io/jobs/10570548-applied-scientist-ii-aws-neuron-science-core-algorithm-at-amazon)
Deliverable
production ML models
Required skills
Deep learning model architecture design, deep learning training and optimization, model pruning, low-precision arithmetic (MXFP8/MXFP4), distributed systems, numerical optimization, Python, C++, Java
Preferred skills
High-performance computing, parallel computing, RL post-training, kernel generation, professional software development
Technologies
PyTorch, JAX, Trainium, Inferentia, AWS Neuron
Responsibilities
Own scientific problems end-to-end from research to production, develop production-quality code for large-scale training systems, partner with engineering and hardware teams to inform product stack, mentor scientists and interns, publish research at top venues (ICLR, NeurIPS, ICML, MLSys).