Latest AI JobsBrowse Jobs
B

Software Engineer - GPU Kernels

BasetenSan FranciscoEPD
Apply on company site
Salary
$180K - $360K USD
Job type
Full-time
Work mode
Remote
Posted
17 Jul 2025
Status
Open · checked 7d ago

About this role

Baseten is currently hiring a Software Engineer - GPU Kernels to join the team in San Francisco, with remote options available. This role is ideal for an engineer passionate about low-level optimization and high-impact machine learning infrastructure.

Role Overview

Baseten is looking for a GPU Kernel Engineer to focus on the cutting edge of AI acceleration. In this position, your code will directly affect the performance of advanced machine learning models. You will build the foundation for modern artificial intelligence workloads by optimizing computations down to the microsecond. The environment is fast-paced and intellectually stimulating, offering strong growth potential for professionals dedicated to low-level optimization.

Key Responsibilities

Design and develop high-performance GPU kernels for essential machine learning operations like mixture-of-experts routing, attention mechanisms, and matrix multiplications.
Write and refine code utilizing PTX assembly, CUDA, and hardware-specific methods.
Apply advanced performance strategies, including tensor core acceleration, memory coalescing, compute and memory overlap, and warp-level programming.
Introduce modern features such as sparsity and quantization options like FP4 and FP8.
Detect and eliminate system performance bottlenecks with Torch Profiler, Nsight Compute, and Nsight Systems.
Partner with research groups to transition theoretical developments into production systems.
Participate in internal and open-source GPU libraries, and share technical work at events like AWS re:Invent and NVIDIA GTC.

Requirements

Solid grasp of GPU programming paradigms and architecture, covering thread, block, and grid structures, memory hierarchies, synchronization methods, and race condition prevention.
Proficiency in C++ and GPU performance profiling instruments.
Practical knowledge of memory access optimization, bandwidth enhancement, numerical precision, quantization tactics, modern GPU capabilities like tensor cores and asynchronous operations, and the CUDA C++ API.

Nice to Have

Background in Transformer models along with attention optimization techniques such as Flash Attention.
Acquaintance with GPU kernel libraries including Triton, Cutlass, CUB, and Thrust.
Experience in GEMM tuning and distributed or multi-GPU computing.
Prior research publications or conference talks focused on GPU performance.
Past contributions to open-source GPU initiatives.

Skills and Tools

C++
CUDA and PTX assembly
Nsight Systems, Nsight Compute, and Torch Profiler
AWS and Spark ecosystem familiarity

Salary and Benefits

Competitive yearly salary ranging from $180,000 to $360,000 USD, supplemented by meaningful equity.
Full 100% medical, dental, and vision insurance coverage for U.S. employees and their dependents.
Flexible paid time off alongside a company-wide Winter Break from Christmas Eve through New Year's Day.
Paid parental leave programs.
Carrot-provided fertility and family-building stipend.
Company-facilitated 401(k) plan for U.S. staff.

About the Company

Baseten delivers mission-critical inference solutions for dynamic AI enterprises such as Notion, Cursor, OpenEvidence, Gamma, Abridge, Writer, and Clay. The organization combines applied artificial intelligence research, flexible infrastructure, and developer tooling to help frontier AI businesses push models into production. Backed by Altimeter Capital, Conviction Partners, and Spark Capital, Baseten recently raised a $1.5 billion Series F funding round.

How to Apply

Interested candidates should visit the official Baseten careers page to submit an application and join the team.

Skills and tech stack

AWSSparkC++

Questions candidates ask

What is the salary range for the GPU Kernel Engineer position?

The annual salary ranges from $180,000 to $360,000 USD, alongside meaningful equity.

Is this position remote?

Yes, remote work options are available for this role based out of San Francisco.

What technical skills are required for the role?

Applicants must have a strong understanding of GPU architecture, C++ proficiency, experience with the CUDA C++ API, and knowledge of GPU performance profiling tools.

What benefits are included for U.S. employees?

U.S. employees receive 100% coverage for medical, dental, and vision insurance, a company-facilitated 401(k), flexible PTO with a winter break, paid parental leave, and a family-building stipend.

How do I apply for this job?

You can apply directly through the official Baseten careers page.

Ready to apply?

The application is on the official Baseten careers page.

Apply on company site