Description

AWS designs some of the most complex custom SoCs in the world — Trainium chips that power massive machine learning training clusters. Our team builds models of these SoCs that are used across the chip development lifecycle: architecture exploration, design verification, and performance analysis. We need a hands-on engineering manager to lead and scale this modeling effort.

You'll own the modeling stack that chip architects, RTL designers, and verification engineers depend on to build correct, high-performance silicon. This is a technical leadership role where you'll drive both the team's execution and its long-term modeling strategy — including expanding into performance modeling to influence architecture decisions earlier in the chip cycle. This isn't a "manage from the sidelines" role — you'll be in the codebase, debugging model issues, and making architecture decisions alongside your team.

What you'll do:

- Lead the team building functional and performance models of SoC subsystems — from individual IP blocks to full-chip and server-level integration
- Own the modeling methodology and architecture: how models are structured, tested, integrated, and validated against RTL and specs
- Drive the expansion into performance modeling - building cycle-approximate or analytical models that inform architecture trade-offs
- Partner with chip architects and RTL teams to ensure models accurately capture hardware behavior and are delivered on schedule
- Work with design verification teams to integrate models into their validation flows
- Scale and develop a team of modeling engineers, setting the bar for technical depth and delivery
- Build the mechanisms (CI, testing, documentation) that let the team move fast without breaking things

Why this role is interesting:

- You'll directly influence how AWS's custom silicon is designed — your models will shape architecture decisions
- The modeling challenges are deep: multi-subsystem SoCs, complex memory hierarchies, custom accelerator datapaths, network-on-chip, rack-level server architecture, performance at scale
- You'll work at the intersection of software engineering and chip architecture, which is a rare and valuable combination
- Small team with outsized impact on AWS's silicon development velocity

No ML background needed. You'll learn the ML accelerator domain on the job. What matters is deep SoC modeling experience and the ability to lead a technical team through complex chip programs.

About the team
More details about Trainium3, our team's latest achievement, as well as some insights into our team culture:
- (https://www.aboutamazon.com/news/aws/trainium-3-ultraserver-faster-ai-training-lower-cost)