Tech skills of tomorrow: Machine learning

By Simon Allardice

The tech landscape is changing fast.

With a seemingly endless number of programs, languages and frameworks at your disposal, it can be difficult to know what tool is a flash in the pan, and what has the power to shape the future of your technology organization. But it doesn’t have to be a mystery.

We asked four experts in software development, data, machine learning and cloud computing about which hard and soft skills companies will need to thrive in the 2020s. Here’s what to pay attention to and invest in as you map out your tech strategy for the next year (and the next decade).

A few years ago, it was still possible to argue whether machine learning would end up becoming a transformative technology in the general business world.

There’s simply no debate anymore.

It can be easy to dismiss certain styles of short-term predictions — like articles that would breathlessly pronounce 2019 (or 2018, or 2017) as “The Year of Machine Learning.” But what we’ve seen in machine learning is a similar progression as with other revolutionary technologies: It may not completely transform your business or life in any given year, but viewed in the long-term, machine learning’s constant, inexorable progress and impact cannot be ignored.

As this field matures, what we’re seeing now is a greater understanding of what skills are truly valuable — allowing us to shift our focus from the specific implementation details of the technologies themselves, and into how we can use them to provide greater understanding and greater insights.

The concerns are less “What technologies are important?” but rather “Now what?”

The state of machine learning

In past years, there were misconceptions about what skills are required for organizations to make inroads in machine learning and data science, leading to a frantic scramble to recruit PhD-level experience without asking if that level of expertise was necessary or even desirable. But just as most organizations don’t need their developers to write database management systems, cryptography libraries or video decoders (but rather programmers who can implement and extend existing platforms and functionality), most organizations don’t need their own unique battery of PhDs in computer science and computational statistics who can write machine learning algorithms.

What’s far more important is developing the interdisciplinary ability to know when and how to use algorithms, treating technical skill in machine learning platforms and frameworks as an accelerator for existing business acumen and domain-specific knowledge. So rather than accreting these insights and expertise only around explicit “data scientist” roles, we see a movement towards a democratization of data science itself — a more widespread uptake of machine learning skills and abilities.

For example, over the last two decades, it’s become common for business users in various roles to use tools like Excel or PowerPoint to construct and present graphs and charts of historical data. It’s so common that it’s now considered a “general business skill,” instead of a specialized capability limited to analysts. But it’s still rare for the average business user to present forward-looking, predictive data — the type of analysis that machine learning can provide. That’s the next step. We’ll hit a new frontier in the democratization of machine learning skills once the use of predictive data is as accessible and common as the use of historical data is right now.

What you need to succeed at machine learning in the future

If you’re still in the early stages of machine learning (as an individual, team or company), it can be confusing to know where to begin. Here are a few focus areas that will help you capitalize on machine learning over the next 10 years:

Technology-agnostic skills

It’s less important you’re an expert on a specific technology like TensorFlow, or a specific machine learning cloud platform, and far more important to be familiar with the general skillset, vocabulary and concepts of machine learning.


Recommending a particular programming language is always a contentious task, but for anyone looking to get started with machine learning and AI, my default suggestion is straightforward: learn Python. Yes, there are other languages popular in the machine learning community, and if you’re already working in an environment or using a specific technology that favors another language — R, for example — then of course use that. Otherwise, Python is never a bad choice, even for those not intending to be hands-on practitioners.

Internal upskilling

One hurdle is the incredible difficulty in hiring and retaining employees with machine learning skills. If you can find external candidates, they lack the internal business knowledge and contextual background to generate insights. Upskilling your existing employees is not a luxury — it’s a necessity.

Data collection

Concentrate on enterprise-level data collection. You can’t analyze what you haven’t tracked.

It’s important to remember that machine learning insights should supplement and inform your people, not serve as a replacement for their knowledge, context and experience. As we see machine learning hold an increasingly important seat at the table of business strategy, we need to be prepared to rapidly upskill our teams — and ourselves.

About the author

Simon Allardice is a staff author at Pluralsight. With over three decades of software development experience, he’s programmed in every discipline, from finance and transportation to nuclear reactors and game development. Prior to joining Pluralsight, Simon was the principal developer author at His first video course released back in 2002.

Since then, his popular courses have been viewed by hundreds of thousands of developers. His current focus is on both the new (the latest iOS and Mac development technologies) and the old (fundamental computer science topics). He obsesses on making complicated subjects accessible, memorable and easier to learn.