Twitch Data Science Interview
In January 2019, over 1.2M users consumed 949M hrs of Twitch Video.
In the same month Twitch had over 55k channels and over 4.6 M active streamers. Their web APIs handle over 50 thousand requests a second on average. The product is nothing short of a phenomenon where millions of hours of content in consumed every single day. It would be an absolute privilege for anyone to learn and build their career around this kind of data.
The size and nature of data at Twitch becomes available once in a lifetime in the career of a Data Scientist.
Twitch data science team has to aggregate data real time from different streams to impact their product in meaningful ways. To make that possible, their data pipeline, collects, cleans, and loads over a billion events a day into their data warehouse. They have also built streaming aggregators to summarize metrics in near real-time for broadcasters.
Interview Process
Their interview process consists of a hiring manager interview, technical/SQL screen and onsite interview. There is a technical/SQL take home test on coderpad before the onsite interview. The SQL take home test has tricky SQL queries which require specific solutions. This test is to make sure one has relevant SQL skills to deal with a gigantic database. Onsite interviews are with specific team members. The onsite interviews are A/B testing, BI and Analytics heavy.
Important Reading
Twitch Data Science team is supported by three pillars — Data Science Research, User Experience Research and data governance. While data scientists sit together and report through one Science organization, they can tie their work back to one of these groups. They use redshift clusters as a backend for their data.
- Twitch Engineering: 5000 ft view into their Org
- A/B Testing Staged Rollouts: Stage Rollout Experiments(Code Included)
- Two-Stage Least Squares For A/B Tests: Twitch A/B Testing
AI/Data Science Related Questions
- How would you create bins for different histograms?
- If you had all possible data and unlimited leeway, what’s the first change you would make to Twitch’s site?
- What metrics would you track in a particular Twitch A/B Test?
- What metrics would you optimize for looking at the results of a Twitch A/B Test?
- How would you pick the outcome of the A/B test?
- Find the number of users logged into Twitch at this moment.
- Based on a given dataset find countries with matching attributes to a games dataset.
- What’s one non-academic thing you’ve done that you’re proud of?
- What is your biggest weakness?
Reflecting on the Questions
Twitch is a hyper growth company. The experience of working here would be very different than other companies doing data science. Hyper growth comes with hyper changes and great learnings. The Data Scientist here would be doing a lot of A/B testing as well as providing ways to improve engagement. A great deal of hyper learning and having a great eye for data can surely land you a job to build the future of live entertainment.
We are building the components of Acing a Data Science Interview at Acing AI. Subscribe to our Acing AI newsletter, I promise not to spam and its FREE!
Thanks for reading! 😊 If you enjoyed it, test how many times can you hit 👏 in 5 seconds. It’s great cardio for your fingers AND will help other people see the story.
The sole motivation of this blog article is to learn about Twitch and its technologies helping people to get into it. All data is sourced from online public sources. I aim to make this a living document, so any updates and suggested changes can always be included. Please provide relevant feedback.
