I am an engineering manager who manages a team of about 20 engineers at Better. My specialties include engineering management, building consumer technology, machine learning, and math. I have managed teams for almost ten years, and I have been doing machine learning for more than ten years.
Head of Engineering
Feb 2015 –
New York City
Joined as one of the first 10 employees.
Working closely with the recruiters, doing 800+ interviews in the first 3 years.
Taking the engineering team from 2 people to 25.
Migrating the legacy tech stack to AWS, Postgres, Python, and microservices.
Writing code together with the team, building many critical components in Python and Node.js.
Managing and mentoring engineers.
Having three managers reporting to me.
Feb 2009 – Feb 2011 (Stockholm),
Oct 2011 – Feb 2015 (New York City)
2011 – 2015 (New York City)
Built up the core of the music recommendation system (80% of the current system is still my algorithms, according to people still at Spotify).
Built an open sourced a data pipeline engine (Luigi) still used by Spotify and many other companies (Capital One, Foursquare, …).
Built an open sourced a library for approximate nearest neighbors (Annoy) still used by Spotify and other companies (Uber, Instacart, …).
Managed the music recommendation team, consisting of about 15 engineers.
Invented and implemented many large scale machine learning algorithms, processing terabytes of data daily.
Awarded three patents on recommender systems.
Brought in a deep learning researcher in 2014 and prototyped a content-based music recommendation system.
Tech lead for a larger group of engineers working on music recommendations, about 25 engineers in total.
Spoke at several conferences and meetups.
Left when Spotify was roughly 1,600 people.
2009 – 2011 (Stockholm)
Joined in 2008 to work on my Master's Thesis when Spotify was roughly 30 people.
Joined full-time in February 2009.
Built and managed a small data team with three people reporting to me
Reported directly to the CTO.
Laid the groundwork for a our data strategy at Spotify.
Ran the first A/B tests, price sensitivity analyses, any many other analyses.
Built a system to simulate and quantify the ad inventory, and identified an opportunity to optimize ad delivery that saved several $MM/year.
Worked closely with the business team and sent them analyses going directly to labels and investors.
Left in Feb 2011, came back in Oct 2011 (after a brief stint at Graham Capital).
High Frequency Trading, Associate
Mar 2011 – Sep 2011
One half of a team managing a $700M+ portfolio.
Implemented high frequency trading strategies in C++ and C#, mostly for commodities and currency futures.
Researched new strategies in Python and R.
Sep 2006 – April 2007
Took time off from university to work at Google for seven months.
Worked on embedded systems for Google's infrastructure.
Worked on large-scale data extraction for Google Maps.
Sep 2003 — Mar 2009
Spent six months at TU Berlin as an exchange student.
Master's thesis ”Implementing a Scalable Music Recommender System“ was awarded best Master's Thesis in Sweden by Naturvetarna (a union).
Represented university at the ACM international collegiate programming championship twice – in San Antonio 2006 and Tokyo 2007.
Won the Nordic programming championship six times.
Sep 2000 — Jun 2003
Won the Swedish programming competition in 2002 and 2003.
Won the Swedish physics competition in 2003.
Placed number 5 in the Swedish mathematics competition in 2003.
Represented Sweden in the International Olympiad in Informatics in Kenosha WI in 2003, placed 8 (gold medal).
Represented Sweden in the International Mathematical Olympiad.
Co-organizer of the NYC Machine Learning meetup which has about 11,000 members.
I write a blog on software engineering which has about 200,000 readers per year.
Authored of a few open source projects: in particular Luigi which is a workflow management system with over 8,000 stars on Github, and Annoy which is a library for approximate nearest neighbors.
Fluent in Swedish, English, and German.