We are a mixture of software, system, and silicon experts using AI every day to deliver the world's most capable and responsive intelligence. We start from the workload. Scaling inference is less about brute force and more about how compute is distributed and memory is interconnected. Inference deserves a better interconnect, not something borrowed from HPC, cloud, or training. If this resonates with you, come join our team.
Hybrid · Montréal
Design and implement communication and execution primitives that enable large-scale AI models to run efficiently across thousands of GPUs. Collaborate across the technology stack, identify performance bottlenecks in distributed training and inference, and benchmark multi-node clusters.