Nextdoor Machine Learning Engineer interview questions

12 Machine Learning Engineers have shared notes from their Nextdoor interviews, with the questions they were asked in each round. That is 41% of the 29 Nextdoor interview notes on Screna AI.

12 Machine Learning Engineer interview notes

Nextdoor·Machine Learning Engineer·Technical Phone Screen

Jul 2026
Nextdoor ML Engineer interview with a topological sort problem that had a tricky lexicographic twist and a follow-up about grouping constraints. The core algorithm wasn't the hard part, the extensions were.
  • Given n items and a list of ordering constraints, return a valid ordering that satisfies all constraints. If multiple valid orderings exist, return the lexicographically smallest one by item ID. If no valid ordering exists due to a cycle, return 'IMPOSSIBLE'. Also analyze time and space complexity.
  • Extend the previous solution to support groups where certain items must appear as contiguous blocks. Both within-group and between-group ordering constraints must be respected. Return any valid sequence or 'IMPOSSIBLE'.

“The base topological sort was fine, but the lexicographic requirement tripped me up for a bit.”

View Post

Nextdoor·Machine Learning Engineer·Recruiter / HR Screen

Jun 2026
HR screen for an ML Engineer role at Nextdoor, covering the usual motivation and behavioral questions. Nothing too surprising but the research paper question caught me slightly off guard.
  • Why do you want to join Nextdoor?
  • What project are you most proud of?
  • Walk me through one of your recent projects.
  • When did you last read a research paper, and what was it about?

“Prepared something about local community impact and how the product sits at an interesting intersection of social graph and real-world geography.”

View Post

Nextdoor·Machine Learning Engineer·Take-home Assignment

Jun 2026
Nextdoor ML Engineer take-home that was basically a full production pipeline crammed into one prompt. Covered everything from data loading to batch inference and they wanted you to explain productionization on top of it. Pretty intense scope for a single assignment.
  • Build a complete end-to-end ML classification pipeline from a CSV dataset, including stratified splits, preprocessing, training a baseline and a stronger model, hyperparameter tuning, evaluation metrics, model persistence, and a batch inference function or CLI. If using a neural network, implement the full training loop explicitly. Also explain your design choices and how you'd productionize this.

“This was a lot.”

View Post

Nextdoor·Machine Learning Engineer·Technical Phone Screen

Jun 2026
Did a coding round for an MLE role at Nextdoor. Just one question, a merge intervals variant, nothing too surprising if you've been grinding leetcode.
  • Given a list of intervals, merge all overlapping ones and return the result.

“Pretty standard if you've seen it before.”

View Post

Nextdoor·Machine Learning Engineer·Technical Phone Screen

Jun 2026
Coding round at Nextdoor for an MLE role. One problem the whole time, a calendar/scheduling variant that looked like a standard intervals question until the weekday wrapping stuff showed up.
  • Given a list of intervals where each interval has a weekday (Monday through Sunday), a start time, and an end time in HH:MM format, merge all overlapping intervals and return the merged result. Intervals can span across consecutive weekdays.

“Recognized it as LC 56 pretty fast, but the weekday dimension threw me off for a minute.”

View Post

Nextdoor·Machine Learning Engineer·Take-home Assignment

Jun 2026
Nextdoor ML engineer take-home where you build a full classification pipeline in Colab on a tabular dataset they provide. Pretty standard end-to-end exercise but they actually want you to talk through your decisions live, not just hand in a notebook with good numbers.
  • Build an end-to-end classification model on a provided tabular dataset in a shared notebook environment. Cover data exploration, preprocessing, modeling, evaluation, and iteration while explaining your decisions out loud.

“The part that tripped me up was the 'communicate as you go' expectation.”

View Post

Nextdoor·Machine Learning Engineer·Technical Phone Screen

Jun 2026
Nextdoor ML Engineer interview that was essentially a full coding exercise: given a raw DataFrame, implement data prep, train a model from scratch with numpy, evaluate it, and then field a bunch of concept questions. No sklearn allowed, which is where things got real.
  • Given a pandas DataFrame with numeric and categorical features and a target column, how would you handle missing values, encode categoricals, split the data, and standardize features without using any ML libraries?
  • Implement logistic regression or linear regression from scratch using numpy, including gradient descent training and a prediction function.
  • What evaluation metrics would you choose for a regression task versus a binary classification task, and how would you detect and address overfitting?
  • Explain the bias-variance tradeoff and how it relates to model complexity.
  • How does L1 regularization differ from L2, and what does each one do to the loss objective?
  • How would you handle class imbalance in a binary classification problem?
  • When does feature scaling matter and why? What happens if you skip it?
  • What is train/validation/test leakage and how do you prevent it during data preparation?

“The no-sklearn constraint hit harder than expected.”

View Post

Nextdoor·Machine Learning Engineer·Technical Phone Screen

Jun 2026
Nextdoor ML Engineer interview that was basically a version string sorting problem dressed up with a bunch of edge cases they clearly wanted you to sweat through. One coding round, algorithm-heavy, with a complexity discussion tacked on at the end.
  • Given up to 100,000 version-like strings (e.g. '1.0', '01.2.0', '2.0.0-alpha', '1.10.3'), implement a comparator that orders them by splitting on '.', comparing numeric segments as integers, treating missing segments as 0, sorting pre-release tags like '-alpha' and '-beta' before the corresponding release, and ignoring numeric build metadata after '+'. Use the comparator to return the sorted list. Also discuss time/space complexity and edge cases like leading zeros, different segment counts, and non-numeric segments.

“I started with the segment splitting and integer parsing, which felt fine, then the pre-release ordering tripped me up for a bit.”

View Post

Nextdoor·Machine Learning Engineer·Technical Phone Screen

Jun 2026
Coding round for an ML Engineer role at Nextdoor. The problem was a feed construction question that felt more like a backend/data engineering puzzle than anything ML-specific, which threw me a bit.
  • Given a list of feed objects each with an id, type (Normal, Video, or Photo), and a numeric score, implement a function that constructs a final ranked feed by sorting objects by score descending, grouping consecutive Photos into batches of up to 3, treating each batch as a single feed item scored by its first photo, then merging everything back into a final descending-score order.

“The sorting part was fine, stable sort by score and move on.”

View Post

Nextdoor·Machine Learning Engineer·Technical Phone Screen

May 2026
Nextdoor ML Engineer interview with a coding question around interval merging. Pretty standard algorithmic problem but the string parsing and day-of-week ordering added some friction I didn't fully anticipate going in.
  • Given a list of meeting time intervals as day-and-time strings in 24-hour format, merge all overlapping intervals per day and return the results sorted by day of week then start time.

“The core merge logic wasn't the hard part.”

View Post