Machine Learning Engineer · Online Assessment · Multiple Types

Pinterest — Machine Learning Engineer ✅ Passed

Level: Intern

Round: Online Assessment · Type: Multiple Types · Difficulty: 7/10 · Duration: 70 min · Interviewer: Neutral

Topics: Machine Learning, Algorithms, Naive Bayes, Decision Trees, Ensemble Learning, Recall, Precision, Confusion Matrix, Gradient Descent, Neural Networks, String Manipulation

Location: Toronto, ON, CA

Interview date: 2026-01-15

Summary

I received an online assessment after submitting my resume and cover letter. The assessment consisted of multiple-choice questions covering ML fundamentals, a coding question, and two ML engineering problems.

The multiple-choice questions tested my knowledge of recall/precision definitions, how to interpret a confusion matrix, gradient-related questions, ensemble learning tradeoffs, and a simple neural network forward pass calculation.

The coding question required me to add parentheses around a plus sign in a string representation of a numeric value to minimize the final calculated result. I used a brute-force approach due to lack of experience with similar problems.

The final two engineering problems involved implementing ML models from scratch. The first one was Bagging decision trees, where I had to implement bootstrapping(), fit(), and predict() without using NumPy. The second one was implementing Naive Bayes from the ground up, including calculating priors and posteriors, and I was allowed to use NumPy.

Details

The online assessment included these types of questions:

• Multiple Choice Questions:

  • Recall and Precision definitions
  • Interpreting confusion matrix
  • Gradient related questions
  • Ensemble learning tradeoffs
  • Neural network forward pass calculation

• Coding Question:

  Given a string representation of a numeric value containing a '+' sign, add parentheses around the '+' sign to minimize the final calculated result. The string does not contain '0'. For example, given "435+122", one possible solution is "4(35+12)2", where the values outside the parentheses are multiplied.

• ML Engineering Questions:

  1. Implement Bagging decision trees with bootstrapping(), fit(), and predict() methods (without using NumPy).
  2. Implement Naive Bayes from scratch, including calculating priors and posteriors (NumPy allowed).