How Data Science Teams Can Succeed at Machine Learning at Enterprise Scale

Start Free Trial
June 13, 2019 by Updated March 30th, 2024

Machine learning is much more than a buzzword — it’s a type of data analysis that can empower many businesses to develop more accurate and more performant technologies. It’s also one of the fastest-growing technology trends. Machine learning adoption among large enterprises reached 58 percent in 2017, according to respondents of a Deloitte survey, and grew to 63 percent in 2018.

To get a sense of how fast businesses are adopting ML, IDC predicts that Artificial Intelligence (AI) spending (which encompasses ML) will grow to $52.2 billion by 2021. This represents a rather astounding Compound Annual Growth Rate (CAGR) of 46.2 percent within the 2016 to 2021 forecast period. This means that spending is increasing by nearly half each year for five years.

The Enterprise Journey to Machine Learning

Whether small or large, in traditional or emerging industries, enterprise data science teams are increasingly using (or looking to use) machine learning to maximize the value of their business data. Inevitably, these teams will experience hurdles and setbacks at various points in their journey.

In our experience working with enterprises, we’ve identified six common challenges that data science teams need to address as they scale up their machine learning operations:

Problem 1: Reconciling disparate interfaces

Problem 2: Resolving environment dependencies

Problem 3: Ensuring close collaboration among all machine learning stakeholders

Problem 4: Building (or renting) adequate machine learning infrastructure

Problem 5: Scaling to meet machine learning requirements

Problem 6: Enabling smooth deployment of machine learning projects

Succeeding at ML at Enterprise Scale

To help data scientists and machine learning engineers who are dealing with any of these six challenges, we have collaborated with O’Reilly Media to bring you “Machine Learning at Enterprise Scale: How Real Practitioners Handle Six Common Challenges.”

This book explores these problems in-depth and offers proven, practical advice for data scientists and machine learning engineers — advice that spans programming interfaces, workspaces, and data processing engines. In particular, we explore how managed cloud services platforms can offer a cost-effective, replicable solution to these problems.

We believe that there are valuable lessons to be learned from experts who have already taken this journey and built successful enterprise-scale machine learning programs. For this report, we interviewed the following real-world data science practitioners to present their unique perspectives and hard-earned advice on handling these common challenges:

  • Nakul Arora, Associate Vice President at Infosys responsible for Infosys’ ML and AI platforms
  • Patrick Hall, Senior Director for Data Science products at H2O
  • Matt Harrison from MetaSnake, a consulting firm that focuses on data science
  • Hussein Mehanna, the Head of AI/ML at Cruise (previously the Director of AI Engineering for Google Cloud)
  • Joao Natali, Director of Data Science at Neustar
  • Jerry Overton, Data Scientist and Technology Fellow at DXC Technology, a global systems integrator
  • Sean Downes, a Senior Data Scientist at a large travel conglomerate

The book was authored by Piero Cinquegrana and Matheen Raza. Piero Cinquegrana is an accomplished data scientist and worked for Qubole as the Senior Product Manager for Qubole’s data science offering. Matheen Raza is a machine learning enthusiast who worked for Qubole as a Product Marketing Manager for data science and machine learning.

Download the free eBook for exclusive insights into the common challenges real data science practitioners face when building enterprise-scale machine learning.

Start Free Trial
Read Introducing a New Presto Scheduler to Improve Cache Reads by Up to 9x in RubiX