Putting the Spotlight on our technical employees
My beat is a blog series that turns the spotlight towards technical employees across various desciplines and roles to showcase what a typical day as a Spotifier consists of.
Tony Jebara: VP of Engineering and Head of Machine Learning
Tony joined Spotify New York last summer as VP of Engineering and Head of Machine Learning. Here, he talks us through his busy working days and explains why he’s glad to be part of the band…
Mornings start with a wake-up call from my three young children. I get them fed, dressed and ready for school. But first, I need an espresso – that’s absolutely necessary if I’m going to get anything done! I don’t really go online until I’ve dropped off the kids around 8.15am, then I hop on the subway and head downtown, checking my email and Slack messages en route.
I have both a vertical and a horizontal role at Spotify – and that means lots of meetings, hangouts and one-to-ones with all the various people I’m working with. The vertical role is heading up the engineering teams that run personalization, which includes the Home page, Search functionality, Voice interaction, and the Programming Platform around content and collections. I have around 115 people reporting to me, mainly in New York, Boston and Stockholm. These teams are very operational, with clearly defined road maps and quarterly delivery targets, so I work with Product Managers and other stakeholders to make sure we’re all staying on track.
Then in addition, I have my horizontal role as Head of Machine Learning, which involves me leading several key work groups across Spotify. One of them is the Machine Learning Tech Strategy work group, made up of representatives from many business units, which decides on the cross-functional points where we invest in Machine Learning and the general capabilities we want to improve on. Right now, for instance, we’re investing in how to get better at online and offline correlation – so that when we’re developing a prototype offline, the work is strongly correlated with what happens when we deploy it online, in the real world. It’s part of a company-wide initiative and just one of the ways we’re improving machine learning at Spotify.
I rarely have much time for lunch, so I usually just head to the salad bar in the office cafeteria and grab something to eat on the go. If I can, I might squeeze in a couple of chats with recruitment or HR people to talk through our strategy on hiring.
As well as my internal meetings, I try to get regular face-to-face time with our partners at places like Google and the SoundBoard council companies that advertise with us. I also speak at conferences to represent the broad machine learning capabilities of Spotify. And every few weeks, I invite an external speaker to come and do a seminar at our office and keep everyone up to speed on what’s going on elsewhere in the industry.
Although I’ve only worked at Spotify a few months, I’ve been really impressed by how open-minded and curious people are here. They’re receptive to technical conversations and want to understand new ways of working. It makes for a high level of innovation – there’s the opportunity to take on new things and a willingness to engage better with our listeners, grow faster and evolve. And there’s also a really collaborative feel here – everyone’s working together to help everyone else succeed. It feels like you’re part of the band. And that’s really refreshing to see, I can tell you.
I try to wrap up my afternoon around 5-5.30pm, so I can be home for my kids’ dinner and hear what’s been happening in their days. But once they’re in bed, I usually spend an hour or two working, sending emails or reading documents. It’s pretty full-on for me during the week – I rarely have a gap in my schedule and if I do, it gets filled very quickly. So on the weekends, I like to pile the kids into the car and head off on a road trip – it’s great to get out of the city and spend time outdoors, just for a change of pace.
Read more from Tony and some of his thoughts on Machine Learning in this Q&A.Tags: machine learning
Published by Spotify Engineering