Pinterest Intern Machine Learning Engineer Interview Experience — 70-Minute OA With 7 ML Multiple Choice and 3 Coding

Pinterest ML Intern OA write-up

There were 10 questions in total, 70 minutes: the first 7 were multiple choice and the last 3 were coding. The multiple choice questions were all ML basics. I wrote down a few that I remember:

• Given several confusion matrices, calculate recall and FPR, and pick the ones that meet the threshold
• Pros and cons of ensemble learning (interpretability, training cost, etc.)
• Which regularization methods push coefficients to 0
• Why bagging helps prevent overfitting
• Hand-calculate the forward pass of a small neural network, with a sigmoid output layer, keeping three decimal places

Coding:

• Find a local maximum in an array
• Implement linear regression with gradient descent by hand (the math formulas are given)
• Implement Gaussian Naive Bayes by hand (the math formulas are given)

Overall it felt okay, but time was a bit tight. If you spend a long time debugging the coding questions, you might run out of time.