I would highly recommend Andrej Karpathy's Hacker's guide to Neural Networks. A step by step guide with little math to cover the basics of NN.
http://karpathy.github.io/neuralnets/
Wow, that surely is thought for programmers.
neuralnetworksanddeeplearning.com is a free online book by Michael Nielsen. I'm reading it right now, and it's wonderful. In the first chapter, he walks you through implementing a basic neural network in Python that works with the MNIST data and uses no libraries (except numpy). He's really good at explaining a lot of the math for a more programmer-ish audience that might not have studied multivariate calculus.
Not sure, I think it's better to use older documentation which is not so centered on RNNs and Convolutional networks. The old docs very well describe a basic net with Backpropagation.