AI Inference Kernel Optimization Engineer

Lever, Inc.; Town of Belgium (WI); Remote; USD 102,000 - 147,000; Full time

Benefits offered by this job

Remote-first

International team

Cutting-edge AI research

Performance-critical infrastructure

Job summary

Lever, Inc. is seeking an AI Research Engineer focused on Kernel & Inference Optimization, based in Belgium with a remote-first, international team. The role spans AI research, low-level engineering, and high-performance model serving across devices from mobile to edge.

You will tackle latency, throughput, and memory challenges, develop GPU kernels (MSL), and advance inference techniques like pruning and quantization. Collaboration with research and engineering teams is essential.

Qualifications

• PhD in NLP/ML or related field with AI research track record preferred.
• Proven expertise in writing compute shaders and MSL.
• Experience with low-level kernel optimization on mobile or constrained devices.
• Track record of improvements in latency, throughput, and memory footprint for AI applications.
• Deep understanding of model-serving architectures and optimization techniques.
• Experience writing GPU kernels for mobile devices.
• End-to-end inference pipelines from optimization to production on constrained hardware.
• Ability to apply empirical research and systematic experimentation to optimize latency and memory.

Responsibilities

• Design and deploy model-serving architectures with high throughput and low latency.
• Develop inference pipelines across mobile and edge platforms.
• Set performance targets for latency, throughput, memory, and reliability.
• Build and run inference benchmarks in simulated and production environments.
• Create datasets and scenarios for evaluating performance under resource constraints.
• Identify bottlenecks and implement batching, memory management, and networking optimizations.
• Develop custom GPU kernels and compute shaders for mobile hardware (MSL).
• Apply pruning, quantization, Flash Attention, KV caching, and speculative decoding to optimize inference.
• Design distributed inference using tensor/pipeline/expert parallelism for large GPUs.
• Collaborate with cross-functional teams to integrate optimized inference frameworks into production.
• Document evaluation methodologies and compare to benchmarks; iterate on optimization strategies.
• Monitor production performance and seek continual improvements in scalability and efficiency.

Skills

MSL expertise

Low-level kernel optimization

Inference optimization

GPU kernels

Distributed inference

Diffusion models knowledge

Vision Transformers knowledge

English communication

Education

PhD in NLP/ML (relevant)

Bachelor's degree in CS or related field

Tools

Metal Shading Language (MSL)

Kernel development tools

Edge devices tooling

Job description

Lever, Inc. is seeking an AI Research Engineer focused on Kernel & Inference Optimization, based in Belgium with a remote-first, international team. The role spans AI research, low-level engineering, and high-performance model serving across devices from mobile to edge.

You will tackle latency, throughput, and memory challenges, develop GPU kernels (MSL), and advance inference techniques like pruning and quantization. Collaboration with research and engineering teams is essential.