Joe Halabi

Joe Halabi is a senior deep learning software engineer at NVIDIA, working on TensorRT Edge-LLM to optimize large language model inference for edge platforms. His work spans GPU kernel development, runtime systems, and power and performance optimization. Previously, he worked on the NVIDIA cuDNN library, developing just-in-time compilation capabilities and optimizing GPU kernels for LLM workloads. He holds a master’s degree in computer science from the University of California, Los Angeles.