Member of Technical Staff - Kernels & GPU Performance
Build and optimize low-level execution primitives for AI inference across various hardware architectures at Gimlet Labs.
Posted by employer 6 months ago
First seen on Joblaze 5 hours ago
Last verified on the company career page 5 hours ago
What you'll build
- Build and optimize kernels for production AI workloads
- Develop execution strategies for accelerator architectures
- Improve memory efficiency and scheduling behavior
- Partner with engineers for performance optimization
- Establish performance engineering standards
Must have
- Strong software engineering fundamentals
- Experience working on performance-critical systems
- Bachelor's degree in a relevant field
Nice to have
- Experience with CUDA, Triton, CUTLASS
- Deep understanding of GPU execution models
- Experience optimizing memory access patterns
- Familiarity with occupancy and latency hiding
- Experience using profiling and performance analysis tools
Requirements
- Education
- Bachelor's degree
Not disclosed in this posting: compensation, years of experience, work arrangement, visa sponsorship.
Joblaze summary
In this role, the Member of Technical Staff focuses on developing and optimizing low-level execution primitives to enhance AI inference performance across various hardware architectures. Key skills include a strong foundation in software engineering and experience with performance-critical systems, particularly in relation to GPU execution models and memory hierarchies. This position is well-suited for individuals with a background in performance optimization and a deep understanding of accelerator programming. Gimlet Labs is in a growth phase, expanding its technology into a production neocloud, which presents opportunities to tackle complex challenges.
Joblaze insights
Quick facts
- What's the tech stack?
- Joblaze extracted these technologies from the posting: CUDA, CUTLASS, GPU, Triton.
- What seniority level is this role?
- Gimlet Labs targets staff-level candidates for this position.
- Is this full-time or contract?
- Full-time for this Member of Technical Staff - Kernels & GPU Performance role at Gimlet Labs.
From the original posting
About the role
As a Member of Technical Staff, you will build and optimize the low-level execution primitives that turn accelerator performance into production inference performance.
Rather than optimizing for one hardware architecture, you will work across accelerators with different execution models, memory hierarchies, capabilities, and software stacks. Your work will shape the latency, throughput, and efficiency Gimlet can achieve across established and emerging hardware architectures.
You will work close to the hardware across kernel implementation, memory access, execution behavior, profiling, and performance validation. You will develop optimizations that account for differences between accelerator architectures and partner with compiler, ML systems, and distributed systems engineers to improve performance across the full execution stack.
What success looks like
In the first 12-18 months, you will:
Build and optimize kernels that improve latency, throughput, and hardware utilization for production AI workloads
Develop execution strategies that unlock performance across both established and emerging accelerator architectures
Improve memory efficiency, scheduling behavior, and execution characteristics across the inference stack
Partner with compiler, runtime, and distributed systems engineers to ensure end-to-end performance optimization
Influence how heterogeneous hardware is deployed and utilized within the next generation of AI infrastructure
Help establish performance engineering standards that shape the future of Gimlet's execution platform
Strong software engineering fundamentals
Experience working on performance-critical systems close to hardware
Comfort reasoning about low-level execution behavior, memory hierarchies, and performance tradeoffs
Bachelor's degree in a relevant field, or an equivalent combination of education, training, and professional experience.
Strong candidates may also have
Experience with CUDA, Triton, CUTLASS, or other accelerator programming models
Deep understanding of GPU execution models (warps/wavefronts, blocks, grids)
Experience optimizing memory access patterns (coalescing, shared memory, cache behavior)
Familiarity with occupancy, latency hiding, and instruction-level parallelism
Experience using profiling and performance analysis tools
Familiarity with multi-GPU or distributed execution is a plus
Solve hard problems.
Own meaningful work.
Build for production.
Help define what’s next.
Standard company text repeated across Gimlet Labs's postings is omitted here.