Find your next role

Description

Amazon Devices is an inventive research and development company that designs high-profile devices like the Kindle family of products, Fire Tablets, Fire TV, Health Wellness, Amazon Echo & Astro products.
This is an exciting opportunity to join Amazon Hardware division in developing state-of-the-art techniques that bring Gen AI on edge for our consumer products. We are looking for exceptional scientists to join our Applied Science team and help develop the next generation of edge models, and optimize them while doing co-designed with custom Machine Learning Hardware based on a revolutionary architecture.


Work hard. Have Fun. Make History.

Key job responsibilities
What will you do?
- Engage in state-of-the-art and innovative research in areas such as Gen AI, model compression, and knowledge distillation
- Contribute to a novel and comprehensive training platform custom-tailored for preparing models for edge applications
- Invent optimization techniques to push the boundaries of deep learning model training
- Derive research approaches from first principles via knowledge of Information Theory, Statistics, Scientific Computing, and Deep Learning Theory
- Create and propose detailed theoretical specifications for novel research ideas and directions, and rigorously justify their correctness
- Train custom Gen AI models that beat the SOTA and paves path for developing production models
- Collaborate closely with compiler engineers, fellow Applied Scientists, Hardware Architects and product teams to build the best ML-centric solutions for our devices by cohesively unifying software and hardware
- Publish in open source and present on Amazon's behalf at key ML conferences - e.g. NeurIPS, ICLR, MLSys

An Ideal candidate would have:
- PhD in quantitative science field, e.g. Applied Mathematics, Statistics, Physics
- Experience with designing novel algorithms via optimization theory and constrained optimization
- Experience with applications of reinforcement learning to GenAI model training
- Experience with training of diffusion models
- Experience with mixture-of-experts (MoE) models