The Tiger Demand Forecasting Solution helps companies rapidly deploy the AI engine at scale to deliver value by generating accurate forecasts and enable better planning and execution. The solution has been enabled for Consumer Product companies across sectors including Food and Beverages, Personal and Household products, Durables and appliances and QSR. This complete white box solution meets the challenges of forecasting in a dynamic environment while addressing complexities around client product hierarchies and business scale. It also provides easy integration into planning tools like SAP, IBP, O9, OM Partners etc.

In this talk, we will discuss the use of Intel oneAPI libraries like oneDNN, oneMKL to optimize the Tiger Demand Forecasting Solution that uses Apache MXNet framework and the benefits we obtained. We will also touch upon some of the challenges faced during this project and how we overcame those challenges to successfully adopt Intel oneAPI libraries.

Learning objectives: We will describe the ease of use of Intel oneAPI libraries to extract the best juice out of any Intel hardware. We will learn how significant performance gains can be obtained with minimal code changes, by making run time changes

Developer will walk away with: Ideas for using oneAPI and related toolkits for AI applications, Simple optimization techniques for AI training and inference