Software Engineer, Metal Runtime (API & Abstractions)

Austinfull timeOn Site

Apply on Employer Site

$100k–$500k

H1B: yes

Tenstorrent Software Engineer on the Metal Runtime team at Tenstorrent designing and evolving high-performance host and device APIs for AI accelerators. Develop abstractions exposing low-level hardware capabilities while maintaining performance and control.

Responsibilities

Design and evolve high-performance host and device APIs for Metal Runtime
Develop intuitive abstractions exposing low-level hardware capabilities
Partner with runtime, kernel, compiler, and software teams to build scalable interfaces
Contribute to runtime implementation, debugging, performance optimization, and API integration
Drive API quality through thoughtful design and usability improvements

Qualification

Comfortable working with concurrencyExperienced in designing cleanThoughtful about balancing performanceCollaborativePartner closely with runtimeContribute to runtime implementationWhat You Will Learn

Required

Strong C/C++ developer with a passion for low-level systems programming and software that runs close to hardware.
Comfortable working with concurrency, processors, memory movement, and performance-critical systems.
Experienced in designing clean, maintainable APIs and abstractions with a focus on developer experience.
Thoughtful about balancing performance, usability, and long-term maintainability in systems design.
Collaborative, curious, and excited to solve challenging problems at the intersection of hardware and software.
Design and evolve high-performance host and device APIs for Tenstorrent's Metal Runtime.
Develop intuitive abstractions that expose low-level hardware capabilities while maintaining performance and control.
Partner closely with runtime, kernel, compiler, and software teams to build scalable, production-ready interfaces.
Contribute to runtime implementation, debugging, performance optimization, and API integration across the software stack.
Drive API quality through thoughtful design, usability improvements, and long-term maintainability.
What You Will Learn
How modern AI runtimes are designed to maximize hardware performance and scalability.