Erik Bernhardsson - Spotify, Recommendation Algorithms & Hiring #65


Erik Bernhardsson is the CTO of Better. Better is on a mission to change the enormous and hopelessly broken mortgage industry. Erik runs the technology team, which consists of roughly 50 engineers.

Before Better, Erik spent six years at Spotify, mostly building the core of the music recommendation system. He started out writing recommendation algorithms and eventually built a team of 20 people to help out with this.

In his spare time, Erik enjoy building open source software, for instance Annoy (6k stars on Github) and Luigi. Erik also co-organizes the NYC Machine Learning meetup, and writes a tech blog which has a few hundred thousand visitors per year.

Earlier in his life Erik used to do a lot of algorithm competitions. He won an IOI gold medal in 2003 and won the Nordic programming competition five times.

If you are enjoying our content, click here to support us!

Episode Summary

“The most impact a CTO can have in many ways is recruiting.”

“At Spotify, for the first year, no one told me who my manager was. No one told me a single thing about what to do. I just did things I thought would be valuable to the business.”

“I’m very happy I chose physics. There are so many things you learn including solid state physics, control theory and complex analysis. When I started doing machine learning and statistics it turned out to be super valuable.”

You’re going to want to find vectors of who wants to listen to what track. Using these vectors you can do all of these recommendation operations. 

—Erik Bernhardsson

Key Milestones

  1. What is it about physics and math that makes them such fertile ground for preparing programmers?
  2. Erik helped build recommendation algorithms and systems for Spotify as a developer and a manager. How do these recommendation systems work and what impact can they have?
  3. What has Erik’s experience working internationally?
  4. Erik made the leap from a successful company to joining a startup as the CTO. How did he know the time was right for such a transition?
  5. What are some of the things Erik looksRange: Why Generalists Triumph in a Specialized World for when hiring people for his team?
  6. What are some concrete interviewing tips?

Additional Resources

Check out Erik’s website/blog -

Develomentor Ep. 17 Jake Mannix – Self-Professed Math Nerd Physicist turned AI Engineer

Develomentor Ep. #44 Aline Lerner – Pro Chef Starts Tech Recruiting Firm,

Range: Why Generalists Triumph in a Specialized World - by David Epstein

A gentle introduction to vectors for machine learning (article) -