Transforming The Enterprise With MLOps

by Ciarán Daly
Article ImageSAN FRANCISCO - Incorporating AI technologies into existing business processes is no small task. While technical solutions continue to proliferate, promising better customer interactions, faster data-driven decisionmaking, and game-changing new insights, there's no clear answers for enterprises looking to get started today.

ParallelM are a Silicon Valley-based company created with the sole mission of helping enterprises automate, scale, and optimize machine learning operations in production. Through MCenter, their flagship product, ParallelM aim to provide enterprises with what they call an 'MLOps' software platform to automate management lifecycles for implementing AI across data scientists, IT operations, and business analysts.

[caption id="attachment_12259" align="alignleft" width="250"]ParallelM executive in Sunnyvale, California, Wednesday, September 13, 2017. (Photo by Paul Sakuma Photography) Nisha Talagala - Co-Founder, CTO, and VP of Engineering @ ParallelM[/caption]

Ahead of ParallelM's participation in The AI Summit San Francisco, we caught up with Co-Founder, CTO, and VP of Engineering, Nisha Talagala, to find out more about how enterprises can start to think about operationalizing machine learning in everyday operations.

Why should large enterprises consider AI? How can they start thinking about their own everyday business problems in relation to AI solutions?

AI has a wide range of business relevant applications, from customer recommendations to process improvements to forecasting. The first step to applying AI is to define the aspect of the business application that can benefit from AI-based predictions (i.e., define the problem). Enterprises should also search out ideal use cases by determining which business problems they have adequate data to apply AI models to.

What does machine learning in production mean in practice for enterprise AI?

In practice, MLOps (machine learning operationalization) for enterprise means the seamless intersection of several areas.

At the execution level, it implies:

  • The automation of machine learning (and other AI) pipelines, algorithms, and integration with existing corporate SDLC (Software Development Lifecycle) practices such as CI/CD (Continuous Integration/Continuous Deployment)
  • The management and orchestration of machine learning applications and their consequent dependent tasks.

At a higher order business level, it also requires practices and mechanisms to optimize business ROI and manage business risk by:

  • ML-specific health and quality detection, explainability of ML applications
  • Full production governance to ensure compliance to business and industry requirements and regulations
  • Business value and KPI correlations to monitor and assure the desired business value.

A successful MLOps practice covers all of these elements and drives seamless collaboration between key stakeholders (Business, Data Science and Operations) across the lifecycle.

Related: How To Not Waste Your Time, Money and Political Capital On AI Moonshots

What are the key obstacles to making AI work for global enterprises?

Difficulty in achieving operationalization is the key challenge preventing global enterprises from realizing value from AI. Existing approaches cannot scale or provide the level of management and visibility needed for companies to truly trust their AI in production applications.

There are also organizational silos within the enterprise that need to be overcome to successfully make AI work. When we work with customers, we find that data science, IT and business teams have not truly discussed their roles in the AI initiatives together. 

How can enterprises best meet the cultural or organisational challenges associated with implementing new AI solutions?

Enterprises are finding that the combined technical, cultural and organizational challenges of successful AI deployment require an upfront definition and implementation of an MLOps practice that incorporates all stages of the AI application lifecycle, and ensures transparency and collaboration across stakeholders.

What does competitive advantage look like in the context of AI?

Competitive advantage comes from being able to productionalize – not just develop –and realize the final ROI value from new AI initiatives faster than the competition, and at a greater scale (more initiatives per quarter, etc.) than the competition.

As told to Ciarán Daly

You can learn more from Nisha and the ParallelM team at The AI Summit San Francisco from September 19-20

Practitioner Portal - for AI practitioners


Hesai and Scale AI open-source LiDAR data set for autonomous car training


Scale claims this is the first time such data has been released with zero restrictions


IBM adds free AI training data sets to Data Asset eXchange


Big Blue has something for you

Practitioner Portal


More EBooks

Upcoming Webinars

More Webinars

Experts in AI

Partner Perspectives

content from our sponsors

Research Reports

More Research Reports


Understanding the advantages of AI chatbots over rule-based chatbots

Infographics archive

Newsletter Sign Up

Sign Up