CRED

Role: Data Scientist

College: Not Specified

Interview Date: October 2023

Interview Type: Off-Campus

Result: Not Specified

Difficulty: Not Specified (Implied Medium-Hard)

Rounds: Technical Interview R1 (only one round detailed)

Topics Asked: ML fundamentals (overfitting-underfitting), projects (data scraping, missing data, loss function, LSTM vs Naive Bayes, OOD performance, ensemble methods, vanishing-exploding gradients), matrix-based programming question

Experience:

This experience details an off-campus Data Science interview at CRED in October 2023.

Technical Interview R1: The interviewer was a senior engineer working in the recommendation space. The round began with introductions, followed by a discussion on ML fundamentals, particularly around overfitting-underfitting. A significant portion (about 20 minutes) was dedicated to the candidate's projects, focusing on the "why" behind decisions, such as data scraping, handling missing data, choice of loss function, and using LSTMs for sentiment analysis versus simpler models like Naive Bayes. Discussions also covered Out-of-Distribution (OOD) performance, baseline model comparisons, and ensemble methods (bagging, boosting, stacking), including how these algorithms improve, their parallelization, and impact on bias/variance. The interview concluded with a matrix-based programming question. A key tip shared was to verbalize thoughts while coding, even for common lines of code. The round was lengthy but covered a broad range of topics.