ODIN: from NumPy to big data, seamlessly

Python has been praised for being a simple language to prototype calculations and algorithms easily. Scientific analysis in Python has been relying (and will rely for many years to come) on the numerical library NumPy which provides a high level, yet efficient array manipulation.

Yet, with the amounts of data that is becoming available and manipulated in the years to come, scientists and programmers (and everybody in between) will need to find easy ways to parallelize algorithms and data using multiple cores provided by a laptop or a cluster offered by super-computers.

ODIN is the new open source package from Enthought to enable programs using NumPy arrays to be used in big data applications. I will present its philosophy, performance and demo how to leverage it to port existing NumPy-based code to parallel computing.