AI Accelerator Hardware Enablement Engineer

Modular; United States; Remote; USD 180,000 - 270,000; Full time

Benefits offered by this job

RSU grants

Healthcare benefits

Team onsites

Job summary

Modular is seeking a Hardware Enablement Engineer to bring up and optimize support for new accelerator architectures across its software stack, from Mojo kernels to model serving. You will work closely with vendor tools and cross-functional teams to enable high performance on emerging hardware.

You will write and optimize Mojo kernels, investigate cross-layer performance issues, and build portability infrastructure while collaborating with hardware partners to ensure robust integration and

Qualifications

• 5+ years in high-performance computing, compiler engineering, accelerator software, systems engineering, or a closely related domain.
• Strong C++ skills and experience contributing to complex, multi-component software systems.
• Hands-on experience with at least one heterogeneous programming model such as CUDA, SYCL, OpenCL, or a comparable accelerator programming environment.
• Understanding of how AI operators are implemented at a low level, demonstrated through experience writing or modifying GPU kernels, custom operators, accelerator code, or working with ML frameworks such as PyTorch at the C++/systems layer.
• Working knowledge of hardware concepts such as memory hierarchy, parallel execution, synchronization, data movement, and the relationship between software performance and the underlying architecture.
• Experience debugging problems that cross software abstraction boundaries and an ability to systematically reason about correctness and performance.

Responsibilities

• Bring up and validate support for new hardware architectures across the Modular software stack, working across kernels, compiler infrastructure, runtime, graph execution, and model serving.
• Write, port, and optimize Mojo kernels for novel accelerator architectures, establishing correctness first and iterating toward strong performance.
• Investigate performance and correctness issues across layers of the stack, including kernel execution, memory movement, compiler lowering, graph decisions, synchronization, and vendor runtime behavior.
• Develop a detailed understanding of new hardware platforms, including their ISA, execution model, memory hierarchy, synchronization mechanisms, compiler constraints, and vendor toolchains.
• Help map important AI operators and workloads onto new architectures, understanding how operations such as matrix multiplication, convolution, reductions, and other model primitives execute efficiently on the target hardware.
• Contribute to portability infrastructure, compiler integration, tooling, testing, and debugging workflows that make it easier to support additional hardware platforms over time.
• Collaborate directly with hardware vendor engineers to understand platform capabilities, investigate issues, build integration tests, and improve end-to-end hardware/software integration.
• Benchmark and profile workloads as hardware support matures, identifying bottlenecks and helping move new platforms from initial correctness toward production-quality performance.
• Share knowledge about emerging architectures with the broader team through technical documentation, demos, design discussions, and engineering write-ups.
• Participate in company events such as onsites and hackathons while contributing to Modular's collaborative and open engineering culture.

Skills

C++

CUDA

OpenCL

SYCL

GPU kernels

AI hardware

Performance debugging

Education

BS in CS/EE

Tools

MLIR/LLVM

Job description

Modular is seeking a Hardware Enablement Engineer to bring up and optimize support for new accelerator architectures across its software stack, from Mojo kernels to model serving. You will work closely with vendor tools and cross-functional teams to enable high performance on emerging hardware.

You will write and optimize Mojo kernels, investigate cross-layer performance issues, and build portability infrastructure while collaborating with hardware partners to ensure robust integration and