Driver Engineer - AI Accelerator Systems (Remote)

Modular, a Qualcomm company; United States; Hybrid; USD 167,000 - 242,000; Full time

Benefits offered by this job

Premier insurance

401k matching

Flexible PTO

Stock options

Job summary

Modular is building a modular AI infrastructure and open‑source stack to simplify AI development and deployment. As a Driver Engineer, you will bridge the compiler, runtime, and silicon, enabling MAX, Mojo, and various accelerators to run efficiently.

You will work across kernels, graphs, and systems, implementing driver abstractions, improving diagnostics, and collaborating with Kernels, Graph Compiler/Runtime, and Serving teams. A strong background in C++ and accelerator APIs is required.

Qualifications

• 3+ years of experience writing high‑performance, low‑latency production systems in C++ (modern C++17/20).
• Strong ownership, lifetime, ABI stability, concurrency and parallelism instincts.
• Solid debugging skills in asynchronous, multi‑device systems (GDB/LLDB, sanitizers).
• Experience with accelerator driver APIs (CUDA Driver API, HIP, Metal, Vulkan).
• Ability to collaborate across Kernels, Graph Compiler/Runtime, and Serving teams.

Responsibilities

• Implement and extend core driver abstractions (Device, Context, Queue, Memory, Function).
• Contribute to multi‑accelerator and multi‑node communication and collectives (NVLink, Infiniband, UCX).
• Improve diagnostics and error reporting across asynchronous execution stack.
• Partner with peer teams to refine surfaces in Mojo and MAX framework.
• Participate in design discussions, code reviews, and maintain high engineering standards.
• Contribute to an open source project and public GitHub repo.

Skills

C++ expert

Concurrency

Debugging skills

Collaborative mindset

Tools

CUDA Driver API

HIP

Metal

Vulkan compute

Job description

Modular is building a modular AI infrastructure and open‑source stack to simplify AI development and deployment. As a Driver Engineer, you will bridge the compiler, runtime, and silicon, enabling MAX, Mojo, and various accelerators to run efficiently.

You will work across kernels, graphs, and systems, implementing driver abstractions, improving diagnostics, and collaborating with Kernels, Graph Compiler/Runtime, and Serving teams. A strong background in C++ and accelerator APIs is required.