CUDA Programmer for AI and High-Performance Computing - Remote

Overview

Join a team built by two professors from Columbia University and Hebrew University who are turning a big academic idea into reality. They’re working on a new technology to make neural networks cheaper and more efficient. The proof-of-concept is ready, and the next step is to bring it to a production level.

In Short

• Strong hands-on experience with CUDA programming.
• Deep understanding of GPU architecture and CUDA internals.
• Experience with implementing algorithms for matrix multiplication and neural network layers.
• Proven ability to debug, optimize, and reverse-engineer CUDA code for performance improvements.
• Passion for exploring the depths of GPU programming and achieving highly efficient solutions.
• Strong communication skills to collaborate with researchers and the technical team.
• At least an upper-intermediate English level.

Requirements

• Implement and optimize CUDA code based on specifications provided by algorithm developers.
• Translate matrix multiplication and neural network layer algorithms into efficient GPU implementations.
• Collaborate with researchers and algorithm developers to ensure accurate implementation of complex mathematical models.
• Dive deep into CUDA internals for optimization, debugging, and troubleshooting.
• Contribute to turning a POC into a robust, production-grade deployment.

Benefits

• People-oriented management without bureaucracy.
• The friendly climate inside the company is confirmed by the frequent comeback of previous employees.
• Flexible working schedule.
• Full financial and legal support for private entrepreneurs.
• Free English classes with native speakers or with Ukrainian teachers (for your choice).
• Dedicated HR.