Ahmed is a seasoned software engineer with an academic background in applied mathematics and computer engineering. His expertise spans machine learning systems, deep learning compilers, and accelerating numerical algorithms. Currently, he is at Modular, working on Mojo’s GPU kernels. Before joining Modular, Ahmed played a key role at Cruise, where he helped build the company’s machine learning compiler and runtime stack from the ground up. Prior to that, he worked at Google Research and Meta AI, where he led multiple projects and made significant contributions to their in-house ML systems. During his time at Google and Meta, Ahmed also made significant contributions to open-source initiatives such as the OpenXLA/IREE Compiler, PyTorch, TensorFlow, and Caffe2.