Our mission is to push machine learning to the extreme-scale. We design and implement exponentially resource-frugal  and  scalable machine learning (ML) algorithms by using randomized  hashing  and  sketching algorithms, suited for modern big-data constraints. Apart from  being  exponentially cheap, the algorithms are embarrassingly parallelizable. The extremely low  resource requirements makes our techniques ideal for IoT devices as well. Furthermore, the algorithms are naturally privacy-preserving as they do not work directly with data attributes and instead only operate on secure hashes or sketches.  See Publications page for more information regarding results and installations.