Applied Scientist II: Hardware‑Aware ML for Trainium

Amazon Science; Cupertino (CA); On-site; USD 172,000 - 222,000; Full time

Benefits offered by this job

RSUs

401(k) matching

Health insurance

Paid time off

Parental leave

Job summary

Amazon Science in Cupertino, CA is seeking an Applied Scientist to advance hardware-aware ML for Trainium and Inferentia. You will own solutions end-to-end from research through production, publishing at top venues, and collaborating with AI/ hardware teams.

The role involves mentoring peers, developing production-grade code in PyTorch or JAX, and shaping the technical roadmap for the AWS Neuron ecosystem. Expect a strong emphasis on ML systems and scalable training.

Qualifications

• PhD or Master's degree and 4+ years in CS, CE, ML or related field.
• Experience in patents or publications at top-tier venues.
• 3+ years building models for business applications.
• Proficiency in Java, C++, Python.
• Experience in state-of-the-art DL models architecture design and training/optimization and pruning.

Responsibilities

• Own scientific problems end-to-end from research through production impact with rigorous evaluation.
• Develop production-quality code in PyTorch or JAX and integrate components into large-scale systems.
• Collaborate with foundation-model, engineering, and hardware teams to inform Trainium/Inferentia integration.
• Mentor scientists and interns; contribute to team goals and roadmap.
• Publish research in top venues (ICLR, NeurIPS, ICML, MLSys) and engage the community.

Skills

Java

C++

Python

Deep learning

Model pruning

Distributed systems

Research & publication

Education

PhD or MS + 4+ yrs exp

Tools

PyTorch

JAX

Job description

Amazon Science in Cupertino, CA is seeking an Applied Scientist to advance hardware-aware ML for Trainium and Inferentia. You will own solutions end-to-end from research through production, publishing at top venues, and collaborating with AI/ hardware teams.

The role involves mentoring peers, developing production-grade code in PyTorch or JAX, and shaping the technical roadmap for the AWS Neuron ecosystem. Expect a strong emphasis on ML systems and scalable training.