Description

Annapurna Labs (our organization within AWS) designs silicon and software that accelerates innovation. Customers choose us to create cloud solutions that solve challenges that were unimaginable a short time ago—even yesterday. Our custom chips, accelerators, and software stacks enable us to take on technical challenges that have never been seen before, and deliver results that help our customers change the world.

As a member of the Annapurna ML/AI Mechanical Thermal Engineering team, you'll own the end-to-end thermal and mechanical architecture for ML/AI accelerator platforms — from initial concept through production deployment and ongoing fleet operations. You'll design mechanical and cooling solutions for some of the highest power-density silicon in the industry, balancing performance, reliability, cost, and operational efficiency at massive scale.

This is a technically challenging role requiring you to operate in ambiguity and at a fast pace. You'll collaborate across silicon design, electrical engineering, firmware, manufacturing, supply chain, and operations teams to deliver platforms that power AWS ML/AI services at global scale.

Key job responsibilities
As a Cloud Hardware Development Engineer (Thermal/Mechanical), you will:

Platform Ownership

Own the complete thermal and mechanical design for ML/AI accelerator platforms — from rack-level infrastructure down to chip packaging, including mechanical packaging, structural integrity, and interconnect systems

Define thermal and mechanical design requirements, establish design targets, and drive cross-functional alignment that enables parallel development across hardware, firmware, and software teams

Deliver production platforms through the full lifecycle: concept, design, analysis, prototyping, validation, manufacturing ramp, and fleet operations

Thermal & Mechanical Design

Design and optimize cooling solutions (air and liquid) for high-power-density ML/AI accelerators

Develop detailed CFD models, compact RC models, and structural FEA for SoC/package thermal analysis and mechanical integrity

Design rack manifolds, cold plates, and data center liquid cooling interfaces for at-scale liquid-cooled deployments

Develop and validate mechanical structures including chassis and enclosures for manufacturability, reliability, and serviceability. Owning tolerance stack-up analysis, GD&T, and DFM for high-volume manufacturing methods (stamping, bending, extrusion, die-casting)

Own mechanical design of integrating high-speed interconnect subsystems including cable cartridges, backplane connectors, and mating interfaces — defining alignment, gatherability, insertion force, and serviceability requirements

Perform structural FEA for shock, vibration, and transportation loads to ensure mechanical integrity across the product lifecycle

Fleet Operations & Reliability

Participate in on-call rotations monitoring fleet thermal telemetry for emergent issues

Perform root cause analysis of thermal and mechanical failures in production, implementing firmware updates, hardware modifications, or operational procedure changes

Cross-Functional Leadership

Drive design standardization

Engage with ODMs and component suppliers to drive design optimization, cost reduction, and supply chain resilience

Lead design reviews, mentor junior engineers, and contribute to the technical direction of the broader organization

Influence chip packaging decisions and establish validation methodologies

About the team
In 2015, Annapurna Labs was acquired by Amazon Web Services (AWS). Since then, we have developed products that power every layer of the AWS cloud, including AWS Nitro, Graviton processors, and custom ML/AI accelerators — Trainium for training and Inferentia for inference — that enable customers to build and run generative AI applications at scale.

The ML/AI MTE team is part of the Annapurna ML/AI hardware development organization. We design and deliver the thermal and mechanical systems for every generation of custom ML/AI accelerator hardware — from chip package through rack-level infrastructure. Our platforms operate at massive scale across AWS data centers globally, with current programs including next-generation Trainium and Inferentia systems featuring liquid cooling at scale.