Accelerating Data Science with GPUs featuring NVIDIA | Online

About this event

At this event, NVIDIA and Dataiku will speak on the tools, skills, and methods for bringing data science into production even faster and more efficiently than ever before.

Tentative Schedule: 6:30pm: Pizza + Beer networking 7:00pm: Productionalization: The Final Frontier with Richard Snow, Analytics Architect at Dataiku 7:30pm: GPU-Accelerated Data Science with RAPIDS with Paul Mahler, Lead Data Scientist at NVIDIA

Talk Abstracts: GPU-Accelerated Data Science with RAPIDS with Paul Mahler, Lead Data Scientist at NVIDIA: The revolution in Deep Learning was made possible by moving neural network computations off of CPUs and on to GPUs. These methods are powerful, but do not often overlap with the day-to-day needs of data scientists working in industry or government.

RAPIDS is an NVIDIA-built suite of open-source, GPU accelerated python packages with an API that looks and feels like Pandas and Scikit-Learn, but operates leaner and faster - even with larger data sets. In this talk, we’ll cover the methodology behind building RAPIDS and how it can speed up work 50x or more with essentially the same code base. I will also cover how RAPIDS can scale from a single GPU to multiple clusters of multiple GPUs, and the future roadmap for the open source project.

Productionalization: The Final Frontier with Richard Snow, Analytics Architect at Dataiku: As organizations become more mature in their data science practices, they've begun to shift from focusing on research and experimentation to what happens and the end of the DS life cycle: putting models into production. In this talk, Richard Snow, Analytics Architect at Dataiku, will demonstrate the three pillars of a good productionalized model: 1) deployment, 2) monitoring, and 3) updating.