Saurav Maheshkar is a data scientist researching attention and representation learning, two of the core ideas behind modern deep learning. He develops and shares his work in the open, publishing code and experiments on GitHub and writing about his findings on his technical blog.