Ahmad Kiswani

Title: Innovations by Ahmad Kiswani in Tensor Processing

Introduction

Ahmad Kiswani is an innovative inventor based in Santa Clara, California. He has made significant contributions to the field of artificial intelligence, particularly in the training of artificial neural networks. His work focuses on improving computational efficiency through advanced data formats.

Latest Patents

Ahmad Kiswani holds a patent for "Tensor processing using low precision format." This invention introduces computer-implemented techniques aimed at enhancing the training of artificial neural networks by utilizing a reduced precision data format, such as float16. The patent details methods for rescaling tensor values before executing matrix operations to avoid overflow and underflow. The innovative data format allows for accurate representation of tensors, ensuring that matrix operations maintain their precision.

Career Highlights

Ahmad Kiswani is currently employed at Nvidia Corporation, a leading company in graphics processing and artificial intelligence technologies. His role involves leveraging his expertise in tensor processing to contribute to cutting-edge advancements in AI.

Collaborations

Ahmad collaborates with notable colleagues, including Boris Ginsburg and Sergei Nikolaev. Their combined efforts in research and development foster an environment of innovation and creativity.

Conclusion

Ahmad Kiswani's work in tensor processing represents a significant advancement in the field of artificial intelligence. His innovative approach to using low precision formats is paving the way for more efficient neural network training.