<img height="1" width="1" style="display:none;" alt="" src="https://px.ads.linkedin.com/collect/?pid=1232938&amp;fmt=gif">
Insights > Blog

Bridging the Gap Between IT and Data Science

By Bryan Gilcrease & Paul Welch | Posted on February 24, 2021 | Posted in Tech We Like

As more and more enterprises are leaning into newer analytics technologies like artificial intelligence (AI) and machine learning (ML), many are hitting a wall when it comes to effectively putting data scientist models into action.

In fact, by some estimates, nearly 90% of models never make it into production.

There are a number of factors behind this high failure rate, including the relative newness of the technologies and a lack of on-hand experience. But more often than not, the main culprit is a disconnect between the data scientists who create analytics models and the IT teams charged with putting those models into production.


What’s causing this disconnect? That common enemy in business known as silos. Specifically:

  1. Data scientists create AI/ML models in isolation
  2. Those models are then handed off to IT for production
  3. IT team is often left having to learn and support new technology to manage these models

This path is often compounded by the fact that most data scientists don’t come from  an IT background, so the models they create can be complicated to make work at best, nearly impossible to configure for production at worst.

Read next: “How to Get More from Data Science"

Breaking down barriers

In recent months, there’s been an increase in enterprises adopting a DevOps approach to AI/ML workloads.

The goal of this approach is to ensure everyone from data scientists to IT are on the same page from the outset of a project all the way through production.

The key to achieving this level of coordination is the creation of cross-functional teams where data scientists work closely with IT and applications developers to provide them with the tools and methodologies that match current software development trends like repositories, containers, and self-service data science environments.

The benefits of this team design are fourfold:

  1. IT is never left in the dark when it comes to what AI/ML models are trying to achieve and how those models will be implemented
  2. Data scientists are able to craft their models in a way that leverages the strengths of the enterprise’s IT infrastructure, which leads to a higher success rate
  3. Application developers are better able to integrate AI/ML models into their development workflows in order to create better products
  4. Promotes best practices in data engineering, data governance, security, and data access methods.

Of course, not every enterprise building out its advanced analytics capabilities has the time or resources available to design cross-functional teams—at least not without disrupting their ongoing operations.

That’s where Redapt can help. Our team of experts can help your organization get started with AI and ML, as well as other tools. We can also offer you guidance on how you can construct your teams in a way that makes your advanced analytics initiatives successful.

To learn more about adopting advanced analytics, or to receive help building out the right teams so your projects effectively reach production, contact one of our experts today.