Latest AI JobsBrowse Jobs
Luma AI logo

Research Scientist / Engineer – Performance Optimization

Luma AIRemote, EUResearch & AI
Apply on company site
Salary
Not listed
Job type
Full-time
Work mode
Remote
Experience
Senior
Posted
8 Oct 2026
Status
Open · checked 7d ago

About this role

This remote role based in the EU suits an expert in GPU optimization and transformer internals who wants to make multimodal models fast. The position centers on profiling and optimizing accelerator, CPU, and GPU code to ensure efficient training and large-scale deployment.

Role Overview

Luma AI seeks a Research Scientist / Engineer for Performance Optimization to work remotely within the EU. The professional in this position focuses on writing kernels and operations that maximize hardware performance without losing model quality.

Key Responsibilities

Profile and optimize GPU, CPU, and accelerator code to achieve minimal latency and maximum utilization
Write high-performance CUDA, Triton, and PyTorch code, creating custom operations as required
Develop fused kernels while leveraging modern hardware features and tensor cores across platforms
Optimize implementations and model architectures for distributed production deployment across multiple nodes
Construct performance analysis tools, monitoring systems, and automation
Implement and research advanced optimization techniques for transformer models

Requirements

Expert-level capability in GPU optimization and Triton or CUDA programming
Strong proficiency in PyTorch, covering custom operations and kernel development
Familiarity with profiling tools such as NVIDIA Nsight, torch profiler, and custom tooling
Deep comprehension of attention mechanisms and transformer architectures

Nice to Have

Background with compilers and exporters including torch.compile, TensorRT, ONNX, and XLA
Experience optimizing inference workloads regarding throughput and latency
Familiarity with kernel fusion techniques and the Triton compiler
Understanding of advanced CUDA optimization and warp-level intrinsics

Skills and Tools

Triton
CUDA
PyTorch
NVIDIA Nsight
Torch profiler
Torch.compile
TensorRT
ONNX
XLA

About the Company

Luma AI aims to construct unified general intelligence capable of generating, understanding, and operating within the physical world. The organization emphasizes that multimodality is essential for intelligence, viewing vision as the critical advancement beyond language models. Luma AI operates as an equal opportunity employer.

How to Apply

Interested candidates should apply on the official company careers page.

Skills and tech stack

Questions candidates ask

Where is this job located?

This position is remote, with a location in the EU.

What core technical skills are required?

The role requires expert-level Triton and CUDA programming, GPU optimization, strong PyTorch skills, proficiency with profiling tools like NVIDIA Nsight, and a deep understanding of transformer architectures.

What will I own in the first 90 days?

The first 30 days involve profiling training and inference paths. Days 30 to 60 focus on shipping an architecture or kernel optimization to improve latency or utilization. Days 60 to 90 center on building monitoring and automation to prevent performance regressions.

What experience is nice to have for this role?

Helpful experience includes working with compilers and exporters such as torch.compile, TensorRT, ONNX, and XLA, optimizing inference workloads, Triton kernel fusion techniques, and knowledge of warp-level intrinsics and advanced CUDA optimization.

What is Luma AI's mission?

Luma AI's mission is to build unified general intelligence that can generate, understand, and operate in the physical world, focusing on multimodality and vision.

Ready to apply?

The application is on the official Luma AI careers page.

Apply on company site