AI SoC Modeling Engineer,Annapurna Labs Machine Learning Accelerators,AWS Job Cupertino area,California USA,Engineering

AI SoC Modeling Engineer, Annapurna Labs Machine Learning Accelerators, AWS

AWS's Trainium and Inferentia chips power the world's largest machine learning clusters. Our team builds C++ models of these custom SoCs that RTL designers, verification engineers, and software teams depend on throughout the silicon development lifecycle. We're looking for a modeling engineer to build and own models that directly impact how our chips are designed, verified, and brought to production.

What you'll do:

Develop and maintain high-fidelity functional model of AI/ML accelerator and its SoC subsystems, including compute engines, memory hierarchies, on-chip interconnects, and data paths — translating architecture specs and RTL behavior into accurate, testable C++ models
Validate model behavior against RTL simulations, emulation platforms, or silicon measurements; debug discrepancies and drive model-to-RTL correlation to high fidelity
Partner with design verification teams to integrate models into pre-silicon validation environments and catch architectural bugs early in the design cycle
Collaborate with architects/micro-architects, RTL design engineers, ML SW engineers, and compiler engineers to evaluate architecture and microarchitecture tradeoffs and help make hardware design decisions
Contribute to cycle-approximate performance model effort enabling architectural exploration ahead of RTL availability, early software development
Quantify system-level tradeoffs across compute, memory bandwidth, networking, and storage to influence reference architectures and long-term silicon strategy
Build and improve modeling infrastructure: simulation frameworks, regression suites, automated correlation checks, and coverage-driven validation flows
Develop modeling methodologies and tools that scale across multiple IP blocks and SoC generations, improving team efficiency and model reuse

Why this role is interesting:

Your models are used to verify silicon before it's built — bugs you catch save months of schedule and millions of dollars
You'll work at the intersection of software engineering and chip design, with deep visibility into how custom ML accelerators are architected
As the team scales, there's a clear path into architectural modeling — using your models to influence chip design decisions, not just validate them
Small team, high ownership, direct impact on AWS's most strategic silicon programs

You will thrive in this role if you:

Have built functional or performance models of SoCs, ASICs, GPUs, CPUs, or IP blocks
Are comfortable working with architectural / design specifications or reference implementations and translating them into C++ or SystemC models
Understand verification concepts and have worked with DV teams or in pre-silicon validation environments
Care about model fidelity and have experience correlating models against RTL or silicon
Are interested in expanding into architectural performance modeling as the team grows
Enjoy working on a small, high-impact team where you own significant pieces of the stack

No ML background needed. You'll learn the ML accelerator domain on the job.

This role can be based in Cupertino, CA or Austin, TX.

Basic Qualifications

Experience programming languages such as C/C++, Python, Java or Perl
2+ years writing functional or performance models of hardware (SoCs, ASICs, GPUs, CPUs, IP blocks)
Familiarity with SoC, CPU, GPU, and/or ASIC architecture and micro-architecture

Preferred Qualifications

2+ years of full software development life cycle, including coding standards, code reviews, source control management, build processes, testing, and operations experience
Experience working with DV teams or integrating models into verification flows
Experience with SystemC or TLM-based modeling
Experience correlating functional models against RTL simulation or emulation
Experience developing or calibrating performance models
Familiarity with Modern C++ (20 and beyond)
Experience with PyTest, Google Test, or similar test frameworks
Experience with multi-threaded simulation

Amazon is an equal opportunity employer and does not discriminate on the basis of protected veteran status, disability, or other legally protected…