Intuit Interview Questions — Page 6

Intuit·Data Scientist·Technical Phone Screen

Jun 2026

Intuit data scientist interview with a pretty gnarly SQL question on subscription churn metrics. The whole thing felt more like a take-home case than a live screen, lots of edge cases baked in and they clearly wanted to see if you'd actually think through the definitions rather than just write a query.

Given monthly end-of-month subscription snapshots, write an ANSI SQL query that computes logo churn count, logo churn rate, gross revenue churn, and net revenue retention for August 2025. Your query must handle users missing from one month via full outer join logic, deduplicate multiple snapshots per user per date using a load timestamp, and treat null MRR as zero.
Using the provided sample data, what are the exact numeric outputs your SQL query would return for logo churn count, logo churn rate, gross revenue churn, and net revenue retention?
How would your churn measurement logic change if you were working with transactional cancellation events instead of monthly snapshots, and what considerations around grace periods and partial-month proration would you need to address?

“This is where I spent most of my time and still felt shaky at the end.”

View Post

Intuit·Machine Learning Engineer·Technical Phone Screen

Jun 2026

Interviewed for an MLE role at Intuit and got a pretty focused section on model serialization and deployment formats. Not the flashiest topic but they went deep on it, deeper than I expected for a screening round.

What model formats have you used for saving and serving models in production?
What's your preferred serialization format for ML models and why?
How do you decide between ONNX and a native framework format like a PyTorch checkpoint or a SavedModel?
Walk me through the tradeoffs between Pickle, Joblib, TorchScript, and ONNX for model serialization.
How does your choice of serialization format affect portability, inference latency, and long-term maintainability of a deployed model?

“Pretty open-ended warm-up but I fumbled it by listing formats without connecting them to actual use cases.”

View Post

Intuit·Machine Learning Engineer·Technical Phone Screen

Jun 2026

Phone screen for an MLE role at Intuit. One coding problem, timestamp manipulation, nothing crazy on paper but I fumbled the execution pretty badly.

Given a CSV with employee ID, meeting ID, join time, and leave time, compute each employee's longest continuous working duration in a day, excluding time spent in meetings.

“I kept flip-flopping between pandas and a plain dict approach and that was my downfall.”

View Post

Intuit·Machine Learning Engineer·Technical Phone Screen

Jun 2026

Interviewed for an MLE role at Intuit and the session leaned hard into debugging production ML systems, specifically the gap between offline metrics and what actually happens in production. Not a lot of coding, more of a systems thinking conversation.

Your model performs well on offline evaluation but degrades in production. How do you debug that?
Walk me through how you'd investigate training-serving skew in a deployed model.
If online feature distributions are diverging from what the model saw during training, where do you look first?
Which architectural components do you inspect when tracking down a feature skew issue?

“This is the kind of question that sounds straightforward until you're actually in it.”

View Post

Intuit·Software Engineer·Technical Phone Screen

Jun 2026

Intuit SWE interview with a grid pathfinding problem that sounds approachable until you actually think about it. The safeness factor framing tripped me up a bit before I got the right direction.

Given an n x n grid where some cells contain thieves, find a path from the top-left to the bottom-right corner that maximizes the minimum Manhattan distance from any cell on the path to any thief.

“My first instinct was pure BFS and I wasted a few minutes going down that road before realizing I needed to think about this differently.”

View Post

Intuit·Software Engineer·Technical Phone Screen

Jun 2026

Coding round at Intuit for a software engineer role. One algorithmic problem with a follow-up that pushed into territory I wasn't totally prepared for.

You're given an array of digits representing a non-negative integer (most significant digit first, no leading zeros). Add 1 to the number and return the updated digit array.

“The base case tripped me up less than I expected.”

View Post

Intuit·Software Engineer·Technical Phone Screen

Jul 2026

Interviewed at Intuit for a software engineering role and got a string parsing problem that looked straightforward but had a few gotchas buried in the details.

Given a multiline string representing the output of 'ls -l', parse each line to extract the file size and the filename, then return the filename with the largest file size. Note that filenames can contain spaces.

“The spaces-in-filename part is what gets you.”

View Post

Intuit·Data Scientist·Technical Phone Screen

Jul 2026

Intuit data scientist interview that went deep on A/B testing statistics, like really deep. They wanted full derivations, not just 'run a z-test and move on,' so if you're fuzzy on where the z-test formula actually comes from, this round will find that out fast.

For an A/B test on a binary delivery outcome, which significance test would you use and why is it preferable to a t-test on proportions? Derive the test statistic from first principles starting from the Bernoulli distribution through to the normal approximation, and state the conditions under which that approximation holds.
Using the large-sample A/B test data (control: 920/1000 delivered, variant: 880/1000 delivered), compute the p-value to three significant figures and a 95% confidence interval for the difference in proportions. Show your work.
A small-market rollout gives you much smaller sample sizes: 20 out of 25 delivered in control, 12 out of 18 in variant. How does your test choice change? Walk through whether you'd use chi-square, Fisher's exact, or a z-test, and either compute the exact p-value or explain precisely how you'd get it.
When does a two-sample t-test produce essentially the same result as a z-test for proportions, and when does it break down? Give at least two concrete failure modes.

“This is where I spent most of my mental energy.”

View Post

Intuit·Software Engineer·Technical Phone Screen

Jul 2026

Product Analyst case interview at Intuit, centered on a funnel scorecard for a QuickBooks-style product with 15+ access points and very uneven traffic distribution. One meaty analytical scenario, no fluff.

You're looking at a funnel scorecard (Visit to Application Start to Application Complete to Paid) broken out by 15+ access points with heavily skewed traffic. What three insights would you surface, and how do you prioritize them by impact rather than just by rate?
How would you handle the long tail of low-traffic access points in your analysis? There are 15+ entry points and most have very little traffic.
Based on your insights from the scorecard, what product, experimentation, or data work would you recommend as next steps?

“This is where I tripped up a bit.”

View Post

Intuit·Data Scientist·Technical Phone Screen

Jul 2026

SQL-heavy technical screen for a Data Scientist role at Intuit. The problem was a single multi-part question built around a subscription analytics table, escalating from basic cohort aggregation to YoY growth rates to a paid-only filter variant. Felt more like a take-home in spirit but was done live.

Given a user-level table with signup and subscription dates, write a query that returns, for each calendar month of signups, the total signups, the count who ever subscribed, and the conversion rate.
Extend the query to compute year-over-year subscription rate growth for each month, defined as the percentage change in conversion rate versus the same calendar month in the prior year. Return NULL if the prior year's month isn't in the data.
Now assume the table includes both free and paid subscription types. Recompute the YoY growth metric but treat only paid subscribers as 'converted'. How does your query change?

“The cohort grouping part was fine, DATE_TRUNC on signup_date and a conditional count on subscription_date not being null.”

View Post