Explain LSTM, how could you put it to use for Stock Market predictions. Why would it be a bad idea to build deep learning models for stock predictions if you were to work only with historic stock price data, Are large deep learning models better the smaller models, justify your answers. They also posted an optimisation question that actually did not have a solution and they asked that just to understand how I try and solve problems. They also wanted to evaluate my understanding of Computational Complexity, which was a learning for me from this interview. All of them who I have spoken to about Curvelogics interview experience, they have consistently said the interviews are tough, and lots of learning in the interview. Another positive aspect especially in the final round, they kept trying to understand what I knew rather than what I didn't.