Breaking into machine learning: Connecting the pieces

Breaking into machine learning: Connecting the pieces

Perry Coneybeer
April 16, 2018

Getting into machine learning is a catch-22: engineers need to have experience in order to gain experience, and bright, would-be machine learning engineers have a difficult time finding opportunities to develop their skills. Instrumental’s machine learning lead, Simon Kozlov, has some advice on how to break into this closed-off field.

The way companies approach hiring machine learning engineers is backwards, Simon says. They look for a specific background: several years of prior experience in machine learning is a requirement, and it’s not uncommon for companies to search for people with ten or more years of experience. The field of machine learning is incredibly new, so almost no one besides the field’s pioneers have those ten years of experience. Companies disqualify promising candidates who have a lot of fresh ideas to bring to a discipline that thrives on innovation. If tons of experience is required to get a job in this field, how are machine learning engineers made?

Simon hasn’t always been a machine learning engineer. He started his career in game development, working on 3D graphics and the infrastructure side of the graphics stack. After about a decade, Simon went through what he calls a “midlife crisis for engineers” and began questioning whether what he was doing was the most worthwhile, fulfilling use of his time. He explored a variety of fields, including robotics and blockchain, and decided to pursue machine learning. He quickly discovered the barriers preventing new engineers from entering the field, so he took online courses on Coursera, Udacity, and those published by top universities, and worked on a few projects of his own to build his skillset. It’s fantastic, Simon notes, that world-class resources are available and free to anyone today, and people should be aware of and willing to take advantage of them.

After spending months looking for a company where he could use his new skills, Dropbox gave him a chance to work with their existing small team of machine learning engineers. Since then, Simon has been excelling in the field, and it’s deeply important to Simon that other engineers have opportunities to discover if machine learning might be a field in which they too can excel. He says it’s difficult, but not impossible to break into machine learning, and offers some tips to people who hope to get into the field:

  1. Actively learn everything there is to learn about machine learning. Take online courses, read books and research papers, and be open to new ideas. It’ll take extra effort in addition to your current job, but if you’re truly hungry to learn, it won’t feel like work.
  2. Persist, don’t despair. It takes time to learn new skills, and you will be rejected from some positions you apply to, but remind yourself that you’re always improving.
  3. Make sure to have concrete projects to showcase your skills, like participating in machine learning competitions on platforms like Kaggle or developing an app that uses machine learning. Get them out into the world via blog posts or videos.
  4. Connect with machine learning engineers who have overcome similar obstacles. They remember what it was like to be in your position, and will be happy to help.
  5. Focus on finding the right company that will give you a chance.

That last piece of advice is especially meaningful to Simon. The difficulty of getting into machine learning is not due to a lack of talent, but is a systemic problem in which companies hire from an incredibly narrow pool. That’s why we’re offering a machine learning apprenticeship at Instrumental—it’s an awesome opportunity for an engineer who’s done steps one through three on Simon’s list to gain valuable experience and improve their skills. Our machine learning apprentice will work on solving problems that aren’t yet well-researched or described in the context of deep learning, and the role will require the ability to come up with novel solutions and methods. The engineer will emerge from the apprenticeship with a depth of experience that is difficult to find elsewhere, and a strong mentor.

Simon thinks that becoming a machine learning engineer will be simpler in the future, but that companies need to make a lot of changes in their hiring processes before that can happen. Until then, take Simon’s tips to heart and remember that although breaking into machine learning is difficult, it’s not impossible.

We’re actively seeking a passionate learner for our machine learning apprenticeship. The apprentice will work as part of the Instrumental team on solving complex, real-world problems and will rapidly hone their machine learning skills. If you think you’d be a good fit for the role, or know someone who might be, check out our careers page or reach out to us directly.

Manufacturing Intelligence

Don’t miss a post, subscribe today.