← Back to Directory

Thumbtack interview notes, page 2 of 3

Mid-sized

Most of these Thumbtack write-ups come from Data Scientists (19), then Product Managers (3). The most reported rounds are the technical phone screen (18), the product sense round (2), the cross-functional panel (1) and the system design round (1).

22 interview notes · updated Jul 2026

Thumbtack·Data Scientist·Technical Phone Screen

Jun 2026
Thumbtack DS interview that went deep into marketplace experimentation design. The question was essentially one long, multi-part case about running an A/B test on a pro ranking algorithm, and they expected you to hold the whole thing together across metric definition, randomization tradeoffs, power math, and guardrails. Dense but fair.
  • Thumbtack is updating the algorithm that ranks pros in search results for customer requests. Design an experiment to evaluate the new ranker, covering your primary metric, guardrails with formulas, randomization strategy, power calculations, bias controls, sequential monitoring, and how you'd read out results including cannibalization effects.

“This was basically six questions stapled together and they wanted you to move through all of them without losing the thread.”

View Post

Thumbtack·Data Scientist·Technical Phone Screen

Jun 2026
Thumbtack data scientist interview with a pretty gnarly SQL question involving time zones, window functions, and calendar spine generation. One question, but it had enough moving parts to feel like three.
  • Write a single PostgreSQL query (no temp tables, CTEs only) that produces one row per calendar week in a given date range, showing the week's transaction total and a 3-week rolling sum. Weeks must be defined in America/Los_Angeles local time (Monday start), the calendar must include weeks with zero activity, and the rolling sum must treat missing weeks as zero.

“This one had a lot of layers and I underestimated the timezone piece at first.”

View Post

Thumbtack·Data Scientist·Technical Phone Screen

Jun 2026
Technical screen for a Data Scientist role at Thumbtack that went deep into NLP pipelines and n-gram feature engineering. The questions were specific enough that I felt like I was being quizzed on a paper I'd half-read.
  • Walk through how you'd design a text preprocessing pipeline depending on whether the input comes from typed text, OCR of scanned documents, or speech-to-text output. Cover language handling, normalization, tokenization strategy, stopwords, lemmatization or stemming, and how you'd deal with emojis, URLs, code snippets, and out-of-vocabulary terms.
  • You used unigrams through trigrams in your feature set. Justify that range theoretically and empirically, including the tradeoffs around sparsity, vocabulary explosion, and how the choice interacts differently with linear models versus tree-based or neural models. How did your metrics and feature importances actually shift as you moved from unigrams to bigrams to trigrams?
  • Compare word-level n-grams versus character-level n-grams. In what situations does each approach outperform the other, particularly for noisy text with misspellings or languages with rich morphology?
  • How would you validate this preprocessing and feature pipeline end to end? Describe your train/validation split strategy, how you'd check for data leakage, and how you'd compare this classical approach against a transformer-based tokenizer and embedding.

“This was the opening question and it sprawled fast.”

View Post

Thumbtack·Data Scientist·Technical Phone Screen

Jun 2026
Thumbtack data scientist SQL round, three questions back to back all on the same marketplace schema. Pretty dense for a technical screen, felt like they wanted to see if you could handle window functions and ordered-set aggregates under pressure.
  • Given a marketplace schema with requests, quotes, and categories, compute the 90th percentile of time-to-first-quote (in minutes) per category per calendar week, for requests that received at least one quote.
  • For the 30-day window ending 2025-09-01, find the top 3 pros per category by completed booking count, with ties handled via dense_rank.
  • For quotes created on 2025-08-31 through 2025-09-01, compute each quote's percent_rank of price within its request, where lower price ranks better, and return only quotes in the cheapest 20%. Break ties by lowest quote_id.

“The percentile_cont part wasn't the hard bit.”

View Post

Thumbtack·Data Scientist·Technical Phone Screen

Jun 2026
Thumbtack Data Scientist interview that was basically a full NLP engineering exercise disguised as a coding round. They wanted a TF-IDF implementation from scratch with sparse matrix output, complexity analysis, and unit tests, all without touching sklearn or NLTK. Pretty demanding for what I expected to be a lighter DS screen.
  • Implement TF-IDF from scratch using only Python, NumPy, and SciPy. Your implementation should include a memory-efficient tokenizer with min_df/max_df filtering, smoothed IDF computation, and output a CSR sparse matrix with vocabulary ordered lexicographically.
  • How does your implementation handle out-of-vocabulary tokens at transform time, and how do you support L2 normalization per document row?
  • Add an inverse_transform method that reconstructs the top-k terms for a given document by TF-IDF score.
  • Analyze the time and space complexity of your fit and transform methods.
  • Write a unit test demonstrating correctness of your TF-IDF implementation on a small 3-document corpus that includes repeated terms.

“This was the core of the whole interview and it took basically the full session.”

View Post

Thumbtack·Data Scientist·Technical Phone Screen

Jun 2026
Thumbtack data scientist interview with a pretty involved stats question about regional response rate analysis. One question, but it had a lot of moving parts and I felt like I was being tested on five different things at once.
  • Given job-level data with fields like job category, region, invitations sent, provider responses, and date, how would you determine whether response rates differ by region after controlling for job category mix and time? Walk through your model specification, how you'd handle standard errors, multiple comparisons across many regions, Simpson's paradox detection, and how you'd communicate effect sizes beyond just p-values.

“This was a lot to hold in your head at once.”

View Post

Thumbtack·Product Manager·Onsite - Product Sense / Strategy

Jul 2026
Thumbtack product design interview, one question about building an RV rental app. Pretty open-ended, which I wasn't fully prepared for.
  • Design an app for renting RVs.

“I jumped straight into features and kind of skipped the whole 'who is this for' part, which I regretted about two minutes in.”

View Post

Thumbtack·Data Scientist·Technical Phone Screen

Jul 2026
Thumbtack data scientist interview that leaned heavily into ML fundamentals, specifically unsupervised vs supervised learning tradeoffs and KNN internals. The questions were layered and the interviewer kept pushing deeper, which I was not fully prepared for.
  • How do you decide between clustering and regression when your dataset has only partial labels? Walk through your decision criteria including label availability, what you're trying to optimize, how you'd evaluate results, and how error costs factor in.
  • Compare at least four clustering algorithms across their assumptions, key hyperparameters, scalability, distance metrics, and failure modes. Where does each break down?
  • Give a concrete scenario where you'd pick DBSCAN over K-Means, and another where K-Means is clearly the better call.
  • Explain K-Nearest Neighbors to a non-technical stakeholder using a real-world analogy, then go deeper: how do you choose k, should you weight by distance, what does feature scaling do to it, and how do you deploy it at scale?

“This one tripped me up because I jumped straight to 'use semi-supervised learning' and the interviewer pulled me back to basics.”

View Post

Thumbtack·Data Scientist·Technical Phone Screen

Jul 2026
Technical screen for a Data Scientist role at Thumbtack that went pretty deep into numerical computing fundamentals. Three Python functions, no libraries allowed, and they wanted you to actually know why naive summation breaks down. More math-y than I expected for a DS interview.
  • Implement a function my_min(nums) that returns the minimum value of a list in O(n) time and O(1) space, without using any libraries.
  • Implement my_mean(nums) using a numerically stable online algorithm, and explain why straightforward summation can produce inaccurate results for floating point numbers.
  • Implement my_median(nums) first with an O(n log n) sorting-based approach, then with an O(n) expected-time quickselect solution. Handle both odd and even length lists.
  • How should your implementations handle edge cases like empty lists, None or NaN values in the input, and very large integers or floats?

“Straightforward enough.”

View Post

Thumbtack·Data Scientist·Technical Phone Screen

Jul 2026
Thumbtack data scientist interview with a deep project walkthrough question that covers basically every angle you can imagine. Felt like one question but was really ten questions stapled together.
  • Walk me through your most impactful project from start to finish: how did you define the problem and success metric, what approaches did you consider and rule out, how did you handle feature selection and data leakage, what was your validation strategy offline vs. online, what trade-offs did you make under constraints, and what would you do differently now?

“This is a beast of a question and I did not pace myself well.”

View Post