Machine learning, a subset of artificial intelligence (AI) involving
the study of algorithms and statistical models that systems use to perform tasks by relying on patterns and inference — is one of the highest demand skills in today’s technology job market.
It stands to reason, then, that machine learning engineers are in good place as far as career outlook. These professionals are sophisticated programmers who develop machines and systems that can learn and apply knowledge without specific direction, according to Study.com, an online education platform.
The focus of machine learning engineers goes beyond specifically programming machines to perform specific tasks, Study.com notes. They create programs that allow machines to take actions without being specifically directed to perform the tasks.
Such an engineer might work on the development of a self-driving vehicle, for example, or program services in such a way that they can attempt to identify a specific individual's interests. “From customized news feeds to tailored web searches, machine learning engineers are contributing to the daily lives of many individuals and how they use technology,” Study.com said.
Machine learning engineer job skills
Among the job skills machine learning engineers need to have are computer programming (possibly including knowledge of specific languages such as C++ or Java), math, data analytics, data mining, knowledge of cloud applications, and the ability to communicate well.
Professional social networking site LinkedIn in its Most Promising Jobs listing for 2019 ranked machine learning engineer No. 15. The rankings are based on factors such as high salaries, a significant number of job openings, and year-over-year growth. To determine the best jobs for career opportunity, LinkedIn combined data from millions of member profiles, job openings and salaries.
Job site Indeed reports that machine learning engineer is in its top 10 AI jobs list. Between 2018 and 2019 the site’s analytics team identified the 10 positions with the highest percentage of job descriptions that include the keywords “artificial intelligence” or “machine learning.”
Machine learning engineer topped Indeed’s list for highest paying job in 2019, after coming in third in 2018.
So, what does it take to become a machine learning engineer? To find out, we spoke with Kyle Hamlin, principal machine learning engineer at Sailthru, a provider of personalization software and services.
Education/early life
Hamlin earned a Bachelors of Science degree in cognitive psychology from Colorado State University in 2010 and a Master of Science degree in statistics from Baruch College.
While studying in college and planning his future he had no intention whatsoever of pursuing a career in IT. “My initial career aspiration was to become a counselor/therapist,” Hamlin said.
That’s not to say he lacked any interest in machines or the mechanics behind how technologies work. “I think I’ve always been interested in tinkering with things, whether it was mechanical or digital,” Hamlin said. “So for me, getting into technology was probably always a good fit even if I didn’t initially realize it.”
[ Related: Guided automation for machine learning ]
Job history
One of Hamlin’s first working experiences was as a data modeling intern at data analytics and data warehouse provider Teradata.
“This was an internship so it had to end, but it was foundational to some of the data engineering knowledge I have and use to this day,” he said.
Following the internship Hamlin took a position as a workforce analyst at Sheltering Arms, a children and family services provider. He was responsible for overseeing all metric-based initiatives and driving efficient recruitment, training and management. Other duties included preparing monthly data reports, analyzing metrics to identify top performers internally and externally, performing pension analysis and assessing turnover to reduce attrition rates.
“This job was the first time that I was able to unleash some of my growing analytical capabilities on an organization,” Hamlin said. “Being a nonprofit, they didn’t have much money to invest in software or really anything to aid in making more advanced decisions about their workforce.”
Hamlin designed a reporting system based on ADP data to distribute reports to hundreds of remote managers, in an effort to help them gain insight into their teams. “This was probably one of the largest analytics efforts at the organization to date — and it was run entirely from my old Dell desktop computer,” he said. “Completing a project like that enabled me to sell myself to a company that more closely aligned with my career aspirations.”
Next, Hamlin joined media and marketing services company Mindshare as a digital insights analyst. One of his accomplishments was to stitch together many disparate data sources to automate report generation for hundreds of Unilever brands. “That project helped me sell myself as analytically proficient to interested companies,” he said.
Soon after Hamlin took advantage of an opportunity at marketing and advertising firm Rocket Fuel, where he was a research analyst. “This company is where I believe my career really took off,” he said. “It was, for one, a true technology company, so I was exposed to many like-minded people.”
Hamlin was also able to work with a variety of proprietary machine learning and big data systems. “In the end, the big selling point for me from Rocket Fuel was all the cutting-edge systems I was exposed to and worked with,” he said.
[ Related: Machine learning skills for software engineers ]
In March 2016 Hamlin joined information, data and measurement firm Nielsen as a machine learning engineer. “Coming from Rocket Fuel and having world class big data experience, I was able to spearhead the scaling of a Web application used to manage an online learning system,” he said.
The system was critical for data scientists to gain insight into — and evaluate — their models. “This was a huge selling point, as I had experience building analytical systems, and now I learned how to scale them to production environments,” Hamlin said.
Hamlin joined his current employer, Sailthru, in 2019, and his focus is more aligned with the data engineering space. “I’ve spent my time modernizing existing machine learning systems, while also building the core foundation for a new platform that leverages a unified data lake and the latest in deep learning,” he said.
Memorable moments
The most memorable moment from Hamlin’s career thus far is how it all got started. “After I finished my undergraduate degree, I started applying for graduate schools all over the country — mostly PhD counseling psychology programs,” he said. ‘While I waited for responses from schools, I came to the conclusion that I actually didn’t want to spend the next seven years in school getting a PhD.”
Hamlin figured it would be more valuable to have five years of work experience under his belt after a masters program than zero working experience after a PhD. So he decided to go for a masters in industrial/organizational psychology from Baruch College in New York City.
“During that first semester at Baruch, there were two experiences that changed my career trajectory completely,” Hamlin said. “The first experience was an attempt to create an idea that my best friend and I had: A platform for sharing and mixing clips of music that we created. In all honesty, it was more of a reason for us to stay in touch. But while we both hacked away at our Ruby on Rails app, I realized that I much prefer spending hours debugging and tinkering with code — even if it was aimlessly — over studying for courses” in psychology.
The second experience was the one class that he enjoyed during his first semester in the psychology program, Applied Statistical Analysis for Business Decisions. In that class, students had to design experiments using good experimental design principles to gather data about some hypothesis, and then analyze it using IBM SPSS statistical software.
“The course was hard, but felt familiar to me because a lot of what I did in my undergrad was study experimental design and analyze data that I gathered in SPSS,” Hamlin said. “It was clear to me as I neared the end of the semester that I loved software and I was good at analysis, so I made perhaps the best decision I’ve ever made in my life: I transferred to Baruch’s MS Statistic program.”
Over the next two years Hamlin made it his mission to learn programming with a variety of languages. The field of data science was burgeoning at the time and he knew that if he was skilled in statistical analysis and also had good programming fundamentals, he would be a prime job candidate.
“This memory is very fond to me because it so clearly was a pivotal change in my life and career — and for the better,” Hamlin said. “But at the time, I agonized over whether or not I had the mathematical capabilities to complete a masters in statistics. In the end, although it was very tough, I made the right decision by following my interests over all else.”
Skills and certifications
Learning Python wasn’t necessarily a requirement, “but I was intensely interested in being a good programmer to complement my education in statistics,” Hamlin said. “Learning to program in Python opened many doors for me. It taught me how to think like a software engineer, which I’ve found to be invaluable, before starting to build any analytical, statistical, or machine learning projects.”
Ultimately, the time Hamlin dedicated to learning software engineering paid off, because it helped him excel at delivering stronger solutions for his employers. “As a machine learning engineer whose primary skill set is knowing how to scale machine learning systems to production software environments, I couldn’t have accomplished that without first learning Python.”
Short-term and long-term goals
Hamlin said he has a growing interest in the field of cybersecurity. “I see it as an area that will continue to grow in importance as our society continues to digitize and AI becomes increasingly widespread [and] prevalent,” he said. “Additionally, my best friend and I are actually still at it building software together almost nine years later.
Of course, we weren't always building over those nine years, but we’ve recently created a product to help give fast visual feedback to teams building Web sites. It’s called Volley, and it's one of my goals for the future: build a sustainable business of my own.”
Advice for others seeking a similar path
“Automate everything,” Hamlin said. “A lot of the real-world experience I gained with programming was through trying to program away any task that I had to do more than a few times. I found that even menial tasks required stitching together many sub-tasks and using multiple systems.”
Those automation endeavors made Hamlin quickly realize that code can get messy quickly, and he needed to learn more about good organization and design patterns.