TL;DR
AI Algorithm Engineer: Analyzing and optimizing AI model performance on GPU hardware with an accent on pushing performance toward theoretical hardware limits and validating upcoming GPU designs. Focus on developing new features for the AI software stack in Python and C++ and optimizing GPU kernels.
Location: Onsite in Gdansk, Poland
Company
hirify.global is a global technology corporation focusing on developing and optimizing AI workloads for current and next-generation GPU hardware.
What you will do
- Analyze, profile, and optimize AI models with a focus on GPU performance and efficiency.
- Develop and maintain components of the AI software stack in Python and C++.
- Implement, tune, and optimize GPU kernels using parallel programming languages and frameworks.
- Produce clear and comprehensive documentation for codebases, performance findings, and analysis workflows.
Requirements
- Strong programming skills in C++ and Python.
- Hands-on experience with major AI frameworks (e.g., PyTorch, TensorFlow, OpenVINO).
- Practical, hands-on experience in performance analysis across AI hardware accelerators or GPUs.
- Excellent problem-solving abilities and strong attention to detail.
Nice to have
- Prior experience in kernel programming for GPUs or similar architectures (CUDA, OpenCL).
- Expertise in pre-silicon GPU performance simulation, debugging, and architectural analysis.
- Experience with version control systems such as Git.
Culture & Benefits
- Collaborative, supportive environment with opportunities for professional development.
- Competitive pay and benefits, including stock programs, quarterly bonuses, and health insurance.
- Access to advanced development programs, conferences, and a wide library of courses.
- Flexible working time and benefits supporting employee wellbeing.
- Participation in hirify.global's Great Place to Work program with various groups of enthusiasts.
Будьте осторожны: если вас просят войти в iCloud/Google, прислать код/пароль, запустить код/ПО, не делайте этого - это мошенники. Обязательно жмите "Пожаловаться" или пишите в поддержку. Подробнее в гайде →