Career roadmap: Machine learning engineer

Anyone with “machine learning” in their job title, or even in their sphere of knowledge, is in a good career place these days. People with skills and experience in machine learning are in high demand, and that definitely includes machine learning engineers.

According to the research firm Markets and Markets, the demand for machine learning tools and systems is expected to grow from $1.03 billion in 2016 to $8.81 billion this year, at a compound annual growth rate of 44 percent. Organizations worldwide are adopting machine learning to enhance customer experience and gain a competitive edge in business operations.

IDG

The growth of data is contributing to the drive for more machine learning solutions and skills, the study says. Examples of applications in key verticals include fraud, risk management, customer segmentation, and investment prediction in financial services; image analytics, drug discovery and manufacturing, and personalized treatment in healthcare; inventory planning and cross-channel marketing in retail; predictive maintenance and demand forecasting in manufacturing; and power usage analytics and smart grid management in energy and utilities.

These are just a few of the use cases for machine learning, and engineers are essential to many of these efforts. So, what does a machine learning engineer do?

Machine learning in software development 

In machine learning, individuals design and develop artificial intelligence (AI) algorithms that are capable of learning and making predictions. Machine learning engineers are typically part of a data science team and work closely with data scientists, data analysts, data architects, and others outside of their teams.

According to Study.com, an online education platform, machine learning engineers are advanced programmers who develop machines that can learn and apply knowledge independently. Sophisticated machine learning programs can take action without being directed to perform a given task.

Machine learning engineers need to be skilled in areas such as math, computer programming, and data analytics and data mining. They should be knowledgeable about cloud services and applications. They also must be good communicators and collaborators.

The professional social networking site LinkedIn, as part of its 2022 LinkedIn Jobs on the Rise research, listed “machine learning engineer” as the fourth fastest-growing job title in the United States over the past five years.

[ Also on InfoWorld: AI, machine learning, and deep learning: Everything you need to know. ]

Becoming a machine learning engineer

To find out what’s involved in becoming a machine learning engineer, we spoke with Nicholas Kridler, a data scientist and machine learning engineer at the online styling service provider Dia & Co.

Kridler earned a Bachelor of Science degree in mathematics from the University of Maryland, Baltimore County, and a Master of Science degree in applied mathematics from the University of Colorado, Boulder. 

In graduate school, my focus was computational mathematics and scientific computing,” Kridler says. “I think a career in a tech-related field was my only choice, because I chose to have such a narrow focus in school.”

Early work experiences

When Kridler left graduate school in 2005, he didn’t have a lot of software development experience, so his options were limited. His first job was as an analyst for a small defense contractor called Metron, which produces simulation software.

In October 2006, Kridler joined another defense contractor, Arete Associates, as a research scientist. Arete specializes in developing remote sensing algorithms. “I learned a lot at Arete, including machine learning, software development, and general problem solving with data,” he says.

Kridler left that position at the end of 2012, when data science was beginning to take off, and joined the healthcare technology provider Accretive Health (now R1 RCM) as a senior data scientist. “Accretive was ambitious about incorporating data science, but the tools available at the time made it difficult to make progress,” he says.

Winning the Kaggle competition

While Kridler was employed at Accretive, his boss let him work on a Kaggle competition with a friend from Arete. “The competition involved classifying whale calls from audio data, and felt similar to things I had worked on at Arete,” he says. “We won by a hair, and beat out the deep learning algorithms which were still in their infancy at the time.”

Kridler’s participation and success in Kaggle competitions helped him land a job as a data scientist with the online clothing provider Stitch Fix, in 2014. “Data science was still fairly new, and I felt that a lot of companies were like Accretive in that they were very aspirational about data science but didn’t necessarily have the environment where a team could be successful,” he says.

Stitch Fix seemed much closer to the environment at Arete, where algorithms were core to the business and not just a nice-to-have, Kridler says. He worked as a data scientist at Stitch Fix from 2014 to 2018.

“I was really lucky to have worked there as the company scaled, because I got the opportunity to learn from talented data scientists and data platform engineers,” Kridler says. “I worked closely with the merchandising team developing inventory algorithms. But I also built analytics tools because it helped build a great relationship with the team.”

One of Kridler’s biggest accomplishments at Stitch Fix was developing the Vendor Dash, which allowed brands to access their sales and feedback data. “It provided a lot of value to our brands and was mentioned in the company’s S-1 filing,” he says.

A solid foundation in programming

Kridler left Stitch Fix in 2018 to move to San Diego. In August 2018, he joined Dia & Co., a styling service provider similar to Stitch Fix. As a machine learning engineer, he worked on styling recommendations and led the effort to rebuild a recommendation infrastructure.

“At Dia, I was able to apply the machine learning infrastructure knowledge I developed at Stitch Fix and further develop my skills as an engineer,” Kridler says. Unfortunately, Dia had to cut back, and he spent the next two years working as a data scientist at two companies, before returning to Dia as a lead machine learning engineer.

A combination of school, early work experience, and timing led Kridler to his current role. “There are so many powerful tools that simply didn’t exist when I was in school and when I was starting my career. When I started, I had to work at a much lower level than is required today, and I think that helps me pick up new skills very quickly.”

For example, he learned to program in C and Fortran “and didn’t touch scripting languages like Python until I already had a solid foundation in programming,” Kridler says. “I worked on machine learning algorithms before they were so prevalent, which gave me a bit of a head start.”

A day in the life of a machine learning engineer

The typical workday or workweek varies quite a bit by company, Kridler says. At Stitch Fix, he worked closely with business stakeholders and was responsible for developing a shared roadmap. “This meant frequent meetings to share the current status of initiatives and to plan upcoming tasks,” he says. Slightly more than half his time was spent in meetings or preparing for meetings. The other half was spent on development, whether the deliverable was an algorithm implementation or an analysis. At Dia & Co., his role primarily supports the company’s platforms, which requires fewer stakeholder interactions. “Our stakeholders submit requests that get turned into tickets and we operate much more like a software development team,” he says. “Around 90% of my time is spent writing code or developing algorithms.”

Most memorable career moments

“Winning a competition will always be the most memorable moment, because it opened so many doors for me,” Kridler says. “Hiring for data science has always been difficult, and I felt that I had an advantage because I was able to point to something that clearly showed what I was capable of doing.” Another memorable moment was when Stitch Fix went public, and he was able to see his work mentioned in the company’s S-1 filing. “I feel really fortunate to have been a part of a company that took such a distinct stance on algorithms and data science.”

Skills, certifications, and side projects

I’ve never had to return to school or earn certificates, but I’ve also been fortunate that I’ve been able to learn on the job,” Kridler says. “When I transitioned into data science, I spent a lot of time learning through Kaggle competitions. I have an easier time learning new things if I have a project that lets me apply that knowledge. I’ve written in so many programming languages that it’s not really difficult for me to learn a new language. I don’t pursue any sort of formal training, and rely on publications and documentation to pick up a new skill. I’ve often relied on side projects for expanding my skill set.”

Career goals: Keep building things

Kridler enjoys building things whether, it’s a new algorithm or a company. “I want to be in a position where I get to continue to build things,” he says. “In my current position, it means building upon the infrastructure and expanding the application of the algorithms we have built. In the future, I would like to build upon what Stitch Fix tried to do and show that algorithms are meant to augment, not replace. Whether it’s helping someone make a better decision or removing the need to do the tedious work, I think people focus on the hype of AI without understanding the overall benefit you get from cobbling together lots of little algorithms.”

Inspirations and advice for aspiring engineers

One of Kridler’s inspirations is Katrina Lake, the founder of Stitch Fix, “because she actually wanted to build something different and she did it,” he says. “Christa Stelzmuller, the CTO at Dia & Co., has great ideas about how to use data, and has a great understanding of what does and doesn’t work.”

For developers seeking a similar path to his own, Kridler’s advice is to follow your passion. “I’ve gotten this advice from many people in my career, and you will always have a better time if you are working on something you are passionate about.” It’s also a good idea to “go out and build a lot of things,” he says. “Just like the best way to becoming a good software developer is to write a lot of code, it really helps to have seen a lot of different problems.”