A little off-topic, but hope this would be an another reminder that despite having huge success/hype these days, deep learning/machine learning are just another tools for solving problems, and you can't go really far if you just treat it like a magic framework (i.e. import tensorflow as tf) and fail to understand its underlying principles.