Real-Time Multimodal Inference Architect

Amazon; Seattle (WA); On-site; USD 168,000 - 227,000; Full time

Benefits offered by this job

Health insurance

RSUs

401(k) matching

Paid time off

Parental leave

Job summary

Amazon in Seattle seeks a Senior Inference Engineer to own end-to-end real-time multimodal inference—from research to production—within strict latency budgets.

You will co-design architectures with scientists, optimize streaming serving, and build offline training/evaluation infra to support RL and deployment at scale.

Collaborate across hardware partners to minimize latency and cost while delivering human-like responsiveness.

Qualifications

• 5+ years of professional software development (non-internship).

Responsibilities

• Co-design architectures to surface latency, memory, and cost implications.

Skills

5+ years software development

5+ years programming

4+ years design/architecture

Bachelor's degree in CS or equivalent

2+ years optimizing neural model infer

Transformers/attention knowledge

Education

Bachelor's degree in computer science or equivalent

Tools

Nsight Compute

Nsight Systems

PyTorch

vLLM

CUTLASS

Job description

Amazon in Seattle seeks a Senior Inference Engineer to own end-to-end real-time multimodal inference—from research to production—within strict latency budgets.

You will co-design architectures with scientists, optimize streaming serving, and build offline training/evaluation infra to support RL and deployment at scale.

Collaborate across hardware partners to minimize latency and cost while delivering human-like responsiveness.