CUDA Engineer – Sparse AI Acceleration

Company Description

SparseMindAI pioneers brain-inspired ultra-sparse training algorithms that make high-performance AI models more sustainable and universally deployable. These advanced algorithms enable models to achieve up to 99% parameter reduction without performance losses, significantly decreasing energy costs and improving training and inference efficiency. The innovations by SparseMindAI also enhance model interpretability while contributing to environmental sustainability and advancing the future of AI applications.

CUDA Engineer – Sparse AI Acceleration

We are building a next-generation cloud platform for brain-inspired sparse AI training and inference. Our mission is to make sparse models not only algorithmically superior — but hardware-efficient and production-ready.

We are hiring a CUDA Engineer to design and implement custom GPU kernels enabling semi-structured sparsity (2:4 and 1:4) on NVIDIA GPUs for both training and inference.

Th...