Hamza Laroussi

AI Infrastructure Engineer in Tunis. CUDA kernels, model serving, and making inference actually cheap. Frustrated by hype, motivated by benchmarks