Pushkar Ratnalikar

Title: Innovations by Pushkar Ratnalikar

Introduction

Pushkar Ratnalikar is an accomplished inventor based in San Jose, CA. He has made significant contributions to the field of data optimization, particularly in neural network models. His innovative approach has led to the development of a unique optimization algorithm that enhances the efficiency of data flow graphs.

Latest Patents

Pushkar holds 1 patent for his invention titled "Data Shuffle Optimization." This patent discloses an optimization algorithm designed to reduce the number of data shuffle operators in a data flow graph representing a neural network model. The algorithm identifies single-entry single-exit (SESE) regions within the graph and selects those that consist solely of data shuffle operators. By generating an affine map for each selected SESE region, the algorithm can replace the SESE region with an affine shuffle operator if the implementation cost is lower. This innovative method effectively reduces the total number of operators in the data flow graph, optimizing the neural network model.

Career Highlights

Pushkar Ratnalikar is currently employed at Amazon Technologies, Inc., where he continues to work on cutting-edge technologies. His expertise in data optimization has positioned him as a valuable asset to the company.

Collaborations

He has collaborated with notable colleagues, including Hongbin Zheng and Jie Wang, contributing to various projects that enhance the capabilities of neural networks.

Conclusion

Pushkar Ratnalikar's work in data shuffle optimization exemplifies the impact of innovative thinking in technology. His contributions are paving the way for more efficient neural network models, showcasing the importance of optimization in data processing.