AI DevOps for large-scale 3D Audio experiences
Machine learning is a science that involves learning from data, and deriving inferences based on data. In the industry, this process can get increasingly chaotic and complex with time as the number of models increase. Often there are multiple data scientists working in isolated environments, trying multiple machine learning approaches/experiments, who then go on to produce fragmented results. Furthermore, data products based on machine learning may involve several machine learning components. Under this scenario, it is an incredibly daunting task to track and evaluate experiments to select the best workflows to put into production.
We will describe an end to end framework that tracks code, data and model simultaneously, specifically applied to the field of immersive 3D audio. This framework, is generalizable and can be used to automate evaluation and optimization of the best performing machine learning workflows for large-scale deployment.
Faiyadh Shahid graduated from University of Southern California and Texas A&M University with a specialization in Electrical Engineering, with the highest distinction. His internships at Mathworks and Canon Information and Imaging Institute shaped his passion for software development and rigorous test automation practices . After his Masters, Faiyadh joined EmbodyVR as the second employee in the role of a Research Engineer. While contributing to the development of novel machine learning and signal processing algorithms, Faiyadh built the entire backend infrastructure that currently supports multiple gaming and entertainment clients. He has two patents and two publications to his name. He loves to begin each day with a piece of chocolate cake!