Tower Research

The interviewer started with probability and statistical reasoning. Some of the questions included: Suppose you have two biased coins. How would you determine which coin was used based on a sequence of observed tosses? Derive the Maximum Likelihood Estimator for a Gaussian distribution with unknown mean and variance. How would you optimize a computationally expensive Python ML pipeline? The interviewer also spent time discussing my ML projects. Questions included: Why did you select this particular model over other approaches? How did you establish a baseline? The interview was mathematically intensive. Knowing the definitions of algorithms was not sufficient; the interviewer expected derivations, assumptions, edge cases and intuition behind the methods. The most important preparation areas would be probability, statistics, linear algebra, optimization, classical ML, time-series modelling How did you ensure that your validation results were not affected by data leakage?