I write a lot of software. Sometimes I get around to releasing it in a form that others can use too...
rERPy is a Python toolkit for doing ERP and rERP analysis. (In progress; code works and is solid, needs more docs and conveniences like plotting tools)
ZS is a fast and space-efficient file format for distributing, querying, and archiving very large datasets, such as the Google n-grams. (Ready for use; the code and format are stable and we provide a public release of the 20120701 Google Books n-grams in ZS format).
Patsy is a Python library for describing statistical models (especially linear models, or models that have a linear component) and building design matrices. Patsy brings the convenience of R "formulas" to Python. (Production ready, and in use by a number of other projects)
I'm a core developer on the numpy project, which is the basic library that makes Python useful for numerical computing. Since I have almost no time to actually write code, this mostly means that I pontificate on mailing lists about what I think ought to be done. Sometimes people listen. Perhaps most notably, I was responsible for the addition of the dedicated operator for matrix multiplication to Python's core syntax (coming soon to a Python near you!).
A powerful, accurate, and easy-to-use toolkit for performing colorspace conversions in Python. Particular highlights are support for the state-of-the-art color appearance model CIECAM02, the better-than-CIELAB perceptually-uniform CAM02-UCS space, and color blindness simulations. This talk gives more sense of it. (Status: works great, could use more docs.)
A small tool (built using colorspacious) for visualizing colormaps, and interactively designing new perceptually uniform colormaps. Joint work with Stéfan van der Walt. In particular, this is the tool that was used to design the "viridis" colormap that is the default in Matplotlib 2.0+. For more details, see our talk.
A simple Python wrapper for the SRI language modeling toolkit, which makes it easy to programmatically query smoothed n-gram language models from Python. (Minimal, but stable and working)
A Python wrapper for the CHOLMOD sparse Cholesky factorization library. (Production ready but not currently maintained - looking for someone to adopt!)
Remote display software for X windows programs. Lets you start a program on a remote system and display it on your local system (just like regular X does), BUT unlike regular X, you can then disconnect and reconnect later and your programs will still be there. Also it works much better than regular X over high-latency connections. Really useful for running ipython or matlab on a compute server under tmux or screen, and getting plots to display locally. Especially if your wifi is flakey, or you have long running jobs, or your compute server is on the other side of the world. (Now maintained by others at xpra.org)