159 articles from Spotify Engineering.
At Spotify, change is a constant, and our agile coaches are here to help teams and leaders thrive as we evolve. With agile principles famously part of our DNA, Spotify doesn’t need “transformation coaches,” as they are often known in the industry, but coaches who transform as organizational needs change. In the early 2010s, our first agile coaches established effective delivery and continuous impr
ICYMI Spotify officially began selling enterprise software today: the bundle subscription. Some of these commercial plugins began life as internal tools, built by Spotify developers, for Spotify developers. Other plugins in the bundle are brand new to the Backstage platform. But they all embody our ways of working — our secret sauce for improving developer experience at scale. And now that we’re s
We started developing our vulnerability management platform (VMP) at Spotify in Q2, 2020, and now that we’ve implemented it and use the system in our day-to-day work, we wanted to take a moment to share our journey to help reduce security risks in an efficient and scalable manner.
Needing to deliver faster and more reliably while managing a growing number of contributors and a more complex codebase seems like the fate of every hyper-growth tech company. For platform teams, the challenge is not any different. How can we quickly roll out and increase the adoption of new technologies safely with a growing codebase and organization?
TL;DR As part of Spotify’s goal to , we’ve made climate action one of our five technology pillars and have been integrating sustainability into our everyday engineering practices. This year, climate action was the theme of our company-wide Hack Week, we made climate responsibility part of our official engineering handbook, and we partnered with Thoughtworks to release the Cloud Carbon Footprint pl
TL;DR The core motivation for Spotify’s Client Platform (CliP) team is empowering and unblocking client developers and giving teams the tools they need to ensure a happy and satisfying developer experience (DX). In line with this, we wanted to improve the coding experience for our development teams through infrastructure changes. We conducted research among 318 engineers and learned that:
Back in June, I presented the current state of Spotify’s open source activities to our developers as part of our internal Spotify Engineering Conference, aka SpEC. In the presentation, I go over where we currently are with our open source efforts, what challenges we face, and the vision for where we are going.
Some of the most common questions asked when it comes to work with performance are, How do you convince stakeholders that improving the performance of your project is actually worth the investment? How can you prove that the work is necessary to begin with? Or prove that you have shipped improvements? And what is the impact of certain changes on users in different scenarios?
TL;DR Understanding algorithmic impact is critical to building a platform that serves hundreds of millions of listeners and creators every day. Our approach includes a combination of centralized and distributed efforts, which drives adoption of best practices across the entire organization — from researchers and data scientists to the engineer pushing the code.
Last year, we added support for 36 new languages to our products in one go, for a total of 62 languages. This article describes how we delivered on such an immense localization effort at Spotify. We called the project Scaling Translations.
TL;DR The (SDLC) has always been followed by functional testing to ensure software solutions have all the necessary features and functions. Because of the growing number of cyberattacks, software development stakeholders have been forced to implement security testing as the main track in SDLC to prevent vulnerabilities and flaws in applications or software (assets). A software security assessment
TL;DR Architectural diagrams are the bread and butter of software design and a foundational tool for communication and collaboration on software development. At Spotify, we have an incredibly complex network of thousands of interlinked software systems owned by hundreds of teams, so having a simple way to visualize these connections is essential. While capturing all of our software in one large di
At Spotify, we’ve tried to uphold the tradition of running one major holiday marketing campaign at the end of each year. The campaigns typically run on various platforms such as the Spotify Homepage and the Premium page.
TL;DR With Spotify’s rapid global expansion to more than 180 countries, it’s critical that the infrastructure of user forecasts catch up to business needs. As the business evolves, we need an infrastructure in place that can satisfy the varying needs of stakeholders across the company. To enable fast iterations and facilitate model training, we built a system that allows user forecasts to run both
TL;DR Today we’re announcing the recipients of the first Spotify FOSS Fund. The objective of this fund is to give the independent open source projects we care about a meaningful financial sponsorship to help ensure those projects have the means to continue maintaining and maturing the code we rely on.
Introducing , Spotify’s free open source tool for converting audio into MIDI. Basic Pitch uses machine learning to transcribe the musical notes in a recording. Drop a recording of almost any instrument, including your voice, then get back a MIDI version, just like that. Unlike similar ML models, Basic Pitch is not only versatile and accurate, but also fast and computationally lightweight. It was b
In September 2020, we had the exciting opportunity to partner with with the hopes of creating opportunities for Fellows to: (1) interact with Spotify employees (2) receive training in how to navigate the technical interview process at Spotify and beyond, and (3) become well-versed in the Spotify Internship and Fellowship application process. To help us achieve these goals, we turned to our enginee
TL;DR What is the future of open source going to look like? At Spotify, we think it’s important to play an active role in helping to preserve what’s best about the community today while fostering the changes necessary for it to continue to succeed tomorrow. From increasing the impact of the projects we open source, to starting a fund for independent maintainers, to partnering with other companies
This past year, a few of us who focus on the reliability of (S4A) spent some time taking a look at the collection of incidents directly affecting S4A in 2021. We made some hypotheses, and then went to work analyzing each one to build a set of quantitative and qualitative metrics. We learned a few things, and we hope those insights will help you as well.
TLDR: Spotify is starting a Free and Open Source Software Fund (FOSS Fund) to pay maintainers of independent projects. This new initiative is about giving back to open source developers and is one of the ways we are investing in a more sustainable open source ecosystem for all of us. The fund will start at 100K EUR, with Spotify’s engineers nominating what projects should receive funds and our fun
(formerly ) is a platform that democratizes live audio streams. Creators are able to create rooms and stream live directly to fans who join the room. They can also interact directly with their fans either by text chatting in the app or bringing them up as speakers to interact by voice. But what about when the room ends? A creator could have had an incredible conversation they might want to preserv
In 2017, we launched the , a set of tools that developers could use to programmatically start, stop, and manage Spotify audio playback from the web. This post presents an overview of what you can do with the API, now called the Player API, and some background information about how it came to exist.
TL;DR Within Spotify, we run 20,000 batch data pipelines defined in 1,000+ repositories, owned by 300+ teams — daily. The majority of our pipelines rely on two tools: (for the Python folks) and (for the Java folks). In 2019, the data orchestration team at Spotify decided to move away from these tools. In this post, the team details why the decision was made, and the journey they took to make the t
Tell us more about working on Spotify Wrapped…
Until recently, Search at Spotify relied mostly on term matching. For example, if you type the query “electric cars climate impact”, Elasticsearch will return search results that contain everything that has each of those query words in its indexed metadata (like in the title of a podcast episode).
My early mornings are spent running about after my children – getting them dressed, giving them breakfast and putting on their sunscreen, so they’re all ready for the day ahead. At 8am, I drive my daughter to preschool, while my mom looks after my one-year-old son – she’s been living with us since the start of the pandemic and has been a huge help when it comes to family life and childcare.
TL;DR For years, Spotify’s official engineering blog has been giving you a peek behind the curtain at Spotify R&D. Today, we’re announcing , our new R&D podcast that gives you another view into our tech world. In each episode I’ll talk with Spotify developers about challenging tech problems and give you a firsthand look into what we’re doing, what we’re building, and what we’re nerding out about a
TL;DR: Using the properties of the Poisson bootstrap algorithm and quantile estimators, we have been able to reduce the computational complexity of Poisson bootstrap difference-in-quantiles confidence intervals enough to unlock bootstrap inference for almost arbitrary large samples. At Spotify, we can now easily calculate bootstrap confidence intervals for difference-in-quantiles in A/B tests with
On March 8, we experienced a global outage triggered by issues in a cloud-hosted service discovery system used at Spotify. We were made aware of issues with login at 18:12 UTC / 13:12 ET and started implementing fixes to critical systems at 18:39 UTC / 13:39 ET. This outage affected our users and we apologize for the inconvenience it may have caused. Our service has now fully recovered.
Tell us more about working on Spotify Wrapped…
Tell us more about working on Spotify Wrapped…
At Spotify, we strive to make our apps available to as many people as possible. As mobile developers, that means we want everybody to be able to download our app without hiccups or constraints.
At Spotify, we aim to build and improve our product in a data-informed way. To do that, teams are encouraged to generate and test hypotheses by running experiments and gathering evidence for what works and what doesn’t.
I’m part of the Mambas team at Spotify, which is responsible for all the animation, presentation and personalized data in the Wrapped interactive mobile app. My job is to take the raw data from the backend and enrich it, translate it and format it for different regions before it goes to the frontend mobile engineers. That includes deciding what music to play during data stories and what images and
Building platforms is a hard business. Building platforms for discerning machine learning (ML) practitioners with bespoke needs and a do-it-yourself ethos is even harder. In today’s post, we will give you a peek into how we built ML Home, the internal user interface for Spotify’s Machine Learning Platform, and the product lessons we learned along the way in our quest to entrench it in Spotify’s ML
What does it take to go from an idea for a new playlist, to shipping that playlist to Spotify users all around the world? From inception, to prototyping, to QAing, and finally shipping, releasing a new playlist at Spotify is a long process full of new learnings every time.
Machine learning is what drives personalization on Spotify. We may have a single platform with 381 million different users, but it may actually be more accurate to say there are 381 million individual versions of Spotify, each one filled with different Home pages, playlists, and recommendations. But with a library of over 70 million tracks to thumb through, how do our ML models actually go about m
TL;DR For 2021 Wrapped, we were challenged to visually express a user’s based on how they listened this year. I like to think of it like this: if your music listening data became a person and walked down the street to the neighborhood aura reader, what would that person’s aura look like? That’s the 2021 Wrapped Audio Aura story.
At Spotify, we constantly work on creating the best developer experience possible for our iOS engineers. Improving build times is one of the most common requests for infrastructure teams and, as such, we constantly seek to improve our infrastructure toolchain.
At Spotify, our goal is to connect listeners with creators, and one way we do that is by recommending quality music and podcasts on the Home page. In this two-part blog series, we will talk about the ML models we build and use to recommend diverse and fulfilling content to our listeners, and the lessons we’ve learned from building the ML stack that serves these models.
In of this two-part series, we talked about the challenges we faced with the models we use to recommend content on Home, including:
At Spotify, data rules all. We log a variety of data, from listening history, to results of A/B testing, to page load times so we can analyze and improve the Spotify service. We instrument and log data across every surface that is running Spotify code through a system called the Event Delivery Infrastructure (EDI). Throughout this blog post we make a distinction between the internal users of the E
TL;DR Remember what life was like before smartphones? Remember manually having to sync your computer’s playlists with your iPod every time you added a few songs? One of Spotify’s core products, our mobile app, was designed specifically to leave all of that busywork in the past, changing how we travel with our music forever.
We’ve just open sourced , Spotify’s framework for adding effects to audio in Python. Pedalboard makes it easy to use studio-quality audio effects in your code, rather than just in your digital audio workstation (DAW).
What’s the best way to assess your developers’ experience and performance to discover what they need help with? Is it by measuring something arbitrary, like how many lines of code they’ve written or how many commits they’ve made? Nope. How much useful data are you really getting out of those numbers anyway?
TL;DR Over the years, Spotify’s brand has expanded to encompass a number of products, from mobile apps to to . But sitting at the core is our flagship product, the one that started it all: the desktop app. In of our podcast series, “”, host and Chief R&D Officer Gustav Söderström walks through how the app (and Spotify in general) came to be — and the product lessons you should take away from that
I’m at my best after nine hours’ sleep, so I tend to wake up pretty late, shower and dress as though I’m going into the office. Although I’ve been in Stockholm a while now, I’m still not a fan of the Swedish breakfast of bread and cheese or salami – give me Nutella on toast any day!
TLDR; In episode 08 of our podcast series “”, we share stories and lessons from building and open sourcing , our homegrown developer portal. Hear why a developer-friendly, market-based platform like Backstage could only have been developed at Spotify (where autonomy is prized, not top-down mandates) and why that ends up making Backstage such a flexible fit for other companies, too. and get all our
Last week, Spotify won an award — and we’re not playing it cool. We . We bragged . Our internal Slack is alive with emoji and exclamation points. We’re really very proud of this one.
Team purpose and pride — my team hit those high marks, but it was a long journey to get there from where we started.
When a design system first starts out, the promise of visual consistency glows bright — the ideal product would have only one set of buttons, a unified typography scale, and elements that look the same no matter which designer made the design or which developer programmed them to be real and deployed.
These days, I don’t need an alarm clock – I get woken up bright and early by my daughter! I live in Jersey City, just across the river from the Spotify NY office, and often used to take the ferry into work. But now, there’s no need to commute – instead, I squeeze in an hour of yoga whilst Leya is entertained by her dad.
Search is a well-established functionality across different industries, devices, and applications. When users come to any kind of search, they already have something in mind, whether they come looking for one thing in particular or are open to becoming inspired. Spotify Search is no exception, helping a vast majority of users find joy through search, regardless of the language or method used to se
For the past couple of years, we’ve been on a mission to modernize our Spotify clients by creating one single desktop UI for both the Desktop application and the Web Player.
TLDR; If you’re deploying a service with Kubernetes, you shouldn’t have to use all of your cluster management skills just to perform everyday developer tasks (like seeing which pods are experiencing errors or checking autoscaler limits). Backstage Kubernetes simplifies your deployment workflow by connecting to your existing Kubernetes implementation and aggregating the status of all your deploymen
TLDR: As Backstage turns one, we’re doubling down on our commitment to the open source project and the community we’re building it with.
TL;DR Spotify encourages engineers to become and lean into technologies and skill sets outside of their core specialization. Being a T-shaped developer doesn’t always mean having to learn more code, additional languages, or frameworks. It can be about broadening your outlook on the product-creation process and outside of your discipline. Design Systems Engineering Manager at Spotify Tyce Clee talk
At Spotify we run hundreds of experiments at any given time. Coordinating these experiments, i.e., making sure the right user is receiving the right “treatment” with a population of hundreds of millions of users, poses technical challenges. Adding to the complexity, some of these experiments must be coordinated in the sense that the same user cannot be in two experiments at the same time. These ar
2020 Wrapped is a story of gratitude and resilience. And we’re grateful for the people and teams behind the curtain who built this product experience (??Give them a hand!).
My days begin with an early morning wake-up call from Zev – he comes through at around 5am and we get a couple of dedicated hours of playtime before the rest of the world gets up. It’s fun – this morning, we baked oatmeal cookies. Then once my wife and I are ready, we take the dog, throw Zev in the back carrier and walk all together through the park to his nanny-share. It’s a really lovely way to