Round 1 - Hangman Project Need > 50% accuracy in this round. If you clear the threshold, there is a brief call with a recruiter describing the position. Round 2 - Techincal + behavioural interview with team lead They briefly described the position, quickly went over the CV and asked simple data science questions. Round 3 - with CEO This was not a great experience. The interview was in late afternoon and he was visibly yawning from the start. Asked me to live-code some simple things and was completely poker-faced throughout. Then asked if I would like to spend more time with Hangman and improve it further. The highest accuracy they have ever got was around 72% and asked if I could reach in 60s. So I spent more time, improved my models and finally managed >65%. They called for an interview again with the CEO. Round 4 - CEO again Their submission works through an API and he asked me to change the code to run it locally live on Google Meet. This requires splitting the train set they give into 2 discrete sets, modifying the calls to API to instead redirect to a new function to fetch the "test" word from the split dataset. 45 mins into a 1 hr final interview we were only still talking about the Hangman problem rather than anything else actually useful for the job. I already had another offer by this round so I politely declined saying "Modifying the entire class definitions will take more than the 15 mins we have left on the call" Overall the other people I interacted with tried to sell the position and the company but the interaction with the CEO was not a pleasant experience overall.

1 phone screen, 1 round with general problem on market to portfolio, 2 rounds ( 1 online and 1 onsite ) of Leetcode medium and a resume discussion and general fit with panel . Expect a hangman challenge which I did not get but it can happen .