Tubi interview notes, page 2 of 2

Mid-sized

Screna has 13 Tubi interview write-ups. The most reported round is the technical phone screen (11).

13 interview notes · updated Jul 2026

Tubi·Data Scientist·Technical Phone Screen

Jun 2026

SQL-heavy technical screen for a data scientist role at Tubi. One meaty question about computing user retention metrics from raw event logs, with a bunch of edge cases baked in to see if you'd actually think through the definitions.

Given a raw event log table, write SQL using CTEs to compute new users, retained users, churned users, and net users for a specific target date, broken down by device platform. Definitions are precise: new means first-ever event on that date, retained means active both the day before and on the target date, churned means active the day before but not on the target date, and net equals new plus retained minus churn.

“The definitions sound clean until you start writing the CTEs and realize you need to be careful about what 'first-ever event' means across all of history, not just the two days in scope.”

View Post

Tubi·Machine Learning Engineer·Technical Phone Screen

Jun 2026

Interviewed for an ML Engineer role at Tubitv and got a probability/sampling problem that looked deceptively simple at first glance. The core challenge was implementing weighted random sampling efficiently, which ended up being more of an algorithms question than anything ML-specific.

Design and implement a class that takes an array of positive integer weights and supports a method that returns a random index with probability proportional to its weight, with O(n) construction time and O(log n) per query.

“My first instinct was to just expand the weights into a big array and sample uniformly, which works but blows up memory if weights are huge.”

View Post

Tubi·Machine Learning Engineer·Onsite - System Design / Architecture

Jul 2026

ML system design round at Tubitv for an MLE role, focused entirely on building a thumbnail selection system for a streaming catalog. Pretty deep dive, lots of follow-ups that pushed into territory I wasn't fully prepared for.

Design an end-to-end ML system that selects which thumbnail to show each user for a given streaming title, with the goal of maximizing engagement. Cover problem framing, data, modeling, offline and online evaluation, serving, and production monitoring.
Your training data only has feedback for thumbnails that were actually shown. How do you reason about thumbnails that were never shown to a given user, and how do you evaluate a model that needs to compare all candidates?
If a click-maximizing model starts surfacing sensational or slightly misleading thumbnails that get clicks but result in low watch completion, how do you detect that and adjust the objective?
A brand-new title enters the catalog with five thumbnails and zero engagement history. Walk through exactly what the system does for that title over the first hours and days.
How would you determine whether personalizing thumbnail selection per user is actually worth the added complexity compared to just picking one globally best thumbnail per title? What experiment would you run?

“This was the main question and it ate up most of the session.”

View Post