LinkedIn data scientist interview

The intricacies of gradient topology theory along with coding optimal solutions to dynamic programming problems. A lot of tough questions on the theoretical details of LSTM cells, with white boarding the equations behind the forget gate, and how the memory state learns long term dependencies by controlling the vanishing/exploding gradient. Nah, I’m just fucking with you. I’m drunk.

I have a friend at LinkedIn who legit told me that he asks people the equations of the forget gate of an LSTM in interviews. He’s insane.

I haven’t interviewed at LinkedIn, but I’ve been on a ton of data science interviews and they’re roughly all the same. Some questions about sql, basic probability (bayes theorem, conditional probability...etc), basics around modeling (where is the randomness in random forest and what is its purpose, pros and cons of X vs Logistic Regression), and then techniques for handling common issues, like class imbalance, over fitting, comparing models.

I had my onsite last week for senior data scientist. The process starts with two technical screening, one. SQL(expect two intermediate question with follow up question) screen share and one case study. If you clear both then the on-site is mostly focused around intermediate SQL and R/Python, stats and A/B testing, Statistical modelling (pretty basic ones like linear, logistic and KNN and little bit of random forest), case study (not really any stucture for this). Hope it's helpful.

How was the interview, OP?

How was it?