Towards Optimized Open Dataframe Ecosystem on the Example of OmniSci

This talk will focus on performance optimizations that improved OmniSci in data ingress and dataframe-like operations. We will show how the Arrow library is helpful in data ingress/egress operations and how building unified layered architecture with standardized Pandas-like API, the intermediate layer with DAG, task and IR representations and optimized implementation layer will be beneficial to the ecosystem in general and allows implementers to focus on performance layer to bring the best possible solutions to the end customers.