Put AI to work
June 26-27, 2017: Training
June 27-29, 2017: Tutorials & Conference
New York, NY

Software architectures for building enterprise AI

Qirong Ho (Petuum, Inc.)
4:00pm4:40pm Wednesday, June 28, 2017
Implementing AI
Location: Gramercy East/West Level: Intermediate
Secondary topics:  Hardware, Machine Learning
Average rating: **...
(2.00, 1 rating)

Prerequisite Knowledge

  • Basic familiarity with AI algorithms (deep learning, regression, etc.)
  • Experience with distributed and high-performance computing (useful but not required)

What you'll learn

  • Learn about the inner workings of enterprise AI development software and the AI software architecture and design choices needed to achieve high performance, efficiency, and scalability, support multiple programming languages, and target multiple hardware platforms


Enterprise AI development involves stitching together many open source tools into a pipeline, which presents challenges for AI R&D and engineering teams. Some tools may not support the team’s preferred programming language (Python, R, Scala, etc.), while other tools may become bottlenecks in the pipeline because they do not achieve performance or scalability targets. In the worst case, the tools may not even support the team’s desired hardware configuration (such as distributed computing and GPUs). These challenges can greatly inhibit AI development at enterprises, where brittle glue code, frustrating deployment issues, and poor tool performance even on excellent hardware all pile up to slow teams down.

Petuum, Inc. is building a standardized AI development platform that allows AI teams to use the programming languages they are already comfortable with and that compiles AI code to multiple hardware configurations with efficiency and scalability. Qirong Ho outlines the technical design choices and architecture that enables the Petuum platform to be omnilingual (supporting multiple programming languages) and omnimount (supporting multiple hardware configurations) while still achieving efficiency and scalability that is substantially better than current open source solutions.

Photo of Qirong Ho

Qirong Ho

Petuum, Inc.

Qirong Ho is vice president of technology at Petuum, Inc., an adjunct assistant professor at the Singapore Management University School of Information Systems, and a former principal investigator at A*STAR’s Institute for Infocomm Research. Qirong’s research focuses on distributed cluster software systems for machine learning at big data and big model scales, with a view toward theoretical correctness and performance guarantees, as well as practical needs like robustness, programmability, and usability. Qirong also works on statistical models for large-scale network analysis and social media, including latent space models for visualization, community detection, user personalization, and interest prediction. He is a recipient of the Singapore A*STAR National Science Search Undergraduate and PhD fellowships and the KDD 2015 Doctoral Dissertation Award (runner up).