Sep 23–26, 2019

Large-scale Deep Learning offline platform: Bing's approach

Kai Liu (Microsoft (BING))
2:55pm3:35pm Wednesday, September 25, 2019
Location: 1A 21/22
Secondary topics:  Data, Analytics, and AI Architecture, Deep Learning

Who is this presentation for?

CTO, Director of Data Platform, Data Science Manager

Level

Beginner

Description

We are running a large-scale Deep Learning Offline Platform with tens of thousands of servers. We partition the services around different steps in deep learning project life cycles:
1) Deep Learning Training Service: a set of frameworks and tools for the model training tasks, optimized for resource scheduling, collaboration, and quick iterations.
2) Deep Learning Offline Processing: a set of tools for running a model against massive data to prepare data sets, optimized for throughput and streaming.
3) Deep Learning Vector Service: a set of tools for hosting and running vectors for online/offline use, optimized for fast computation.
4) Deep Learning Inference Service: a set of tools to hosting trained models in offline fashion for model validation and collaboration, optimized for interactivity and cost-efficiency.

In this session, we wish to provide a comprehensive view of a deep-learning offline system that can boost the productivity of data scientist community in the organization.

Prerequisite knowledge

General understanding of key steps in Deep Learning projects

What you'll learn

In this session, we wish to provide a comprehensive view of a deep-learning offline system that can boost the productivity of data scientist community in the organization.
Photo of Kai Liu

Kai Liu

Microsoft (BING)

Kai Liu is a Senior Program Manager in AI and Research group of Microsoft. He has 7 years of experience in data driven engineering, big data platform and AI infrastructure for Office product families. He led his team to create a service health portal for SharePoint Online, inject a distributed log collection and storage system for Exchange Online, publish curated data sets and key business metrics and enable sub-hour experimentations in Office 365. Currently he is working on the AI and Deep Learning infrastructure for large scale enterprise data under compliance obligations.

Leave a Comment or Question

Help us make this conference the best it can be for you. Have questions you'd like this speaker to address? Suggestions for issues that deserve extra attention? Feedback that you'd like to share with the speaker and other attendees?

Join the conversation here (requires login)

Contact us

confreg@oreilly.com

For conference registration information and customer service

partners@oreilly.com

For more information on community discounts and trade opportunities with O’Reilly conferences

strataconf@oreilly.com

For information on exhibiting or sponsoring a conference

Contact list

View a complete list of Strata Data Conference contacts