Skip to contentSkip to footer

Learning Developer Interview Questions

Salaries
Interviews

14,646 learning developer interview questions shared by candidates

Reverse the string Several ML questions - what is bias and variance, what is the difference between classification and regression, What is the effect on the output feature maps of the convolution layer that follows a max pooling layer. Why do we want a maxpooling layer. When to use a confusion matrix. What is the difference between precision, recall and F1. what is focal loss, what is dice loss. ML System Design (Facial recognition system) - think about clustering, dimensionality reduction, database storage, approximate nearest neighbor search, reducing database lookup time with horizontal scaling. Also think about feature/concept drift detection and rectification. Talk briefly about knowledge distillation, model quantization and pruning. Data engineering question - how would you remove duplicate images? Hash the images using a fingerprinting algorithm using DFT. Similar to how shazam's audio fingerprinting works. Now you have reduced the problem to removing duplicate fingerprints. If you type "image hashing" in google, a lot of academic papers with noise robust algorithms will pop up.
avatar

Machine Learning Engineer

Interviewed at Carnegie Mellon University

4.4★
Jul 18, 2022

Reverse the string Several ML questions - what is bias and variance, what is the difference between classification and regression, What is the effect on the output feature maps of the convolution layer that follows a max pooling layer. Why do we want a maxpooling layer. When to use a confusion matrix. What is the difference between precision, recall and F1. what is focal loss, what is dice loss. ML System Design (Facial recognition system) - think about clustering, dimensionality reduction, database storage, approximate nearest neighbor search, reducing database lookup time with horizontal scaling. Also think about feature/concept drift detection and rectification. Talk briefly about knowledge distillation, model quantization and pruning. Data engineering question - how would you remove duplicate images? Hash the images using a fingerprinting algorithm using DFT. Similar to how shazam's audio fingerprinting works. Now you have reduced the problem to removing duplicate fingerprints. If you type "image hashing" in google, a lot of academic papers with noise robust algorithms will pop up.

Viewing 3261 - 3270 interview questions

Glassdoor has 14,646 interview questions and reports from Learning developer interviews. Prepare for your interview. Get hired. Love your job.