Introduction
Hi, I am algorithm engineer with a background in physics and signal processing. Coding is a hobby of mine. I like to contribute fixes and improvements to projects that I use frequently, such as the numpy / scipy / matplotlib stack. My ongoing pet project is scikitguess. It's a collection of linearized optimization routines that started as a translation and python implementation of Jean Jacquelin's paper, which you can find in the repo.
Relevant Links
My Repos
Here is a selection of repos with a note about each one:

scikitguess [PyPi] [RTD] See above.

haggis [PyPi] [RTD] Originally made to support imprint, this library has a whole bunch of routines I find useful on a regular basis for all kinds of projects.

imprint [PyPi] [RTD] A publication system for generating MS Word documents I made while at the Detector Characterization Lab at NASA GSFC. Only really useful in the case where you want to generate the same document over and over, but with different data.

is_integer_ufunc is a foray into the C code of numpy. It contains bit twiddling operations for determining whether an IEEE754 float (any format) is an integer or not. It's becoming pretty clear that a ufunc is inadequate for the purpose, so I am concocting two alternatives to the pure ufunc:
 A numpy function that labels each element with whether it can be stored in an integer of specified bitwidth.
 The number of bits required to store the integer stored in a float, with negative numbers indicating a fractional portion.
See the Stack Exchange section below for the inspiration.

puzzlesolvers [PyPi] [RTD] Another Stack Overflowinspired package. This one is currently just a system for solving a zebra puzzle, which I really enjoyed writing. See the Stack Exchange section for the inspiration.

libpluck never got past the README phase, as it's more of an April fool's library than anything. Developed in conjunction with Ventsy Velev while working on GOESR together. An idea that never made it even this far was a library for generating unit tests with gramatically meaningful, fancy sounding names, all of which pass after much output and fanfare.

BOFH is a small Java program that depends on JTools, and displays a GUI for the excuse generator from Simon Travaglia's BOFH series of articles.

JTools is a small library of Java utilities from many years ago when I coded in Java. It's so old that it was written before Java 7 even existed. There were no streams back in those days. The library has all sorts of tools for I/O, process management, GUI elements, XML parsing and the like.

JToolsextras is an extension of JTools that contains utilities with external dependencies (like TestNG).
Stack Exchange
I've been on Stack Exchange for a while now, and some of the open source work I've done is a direct consequence of my contributions there. The list below maps some of my PRs to the questions that they originated with. In most cases, I added an answer to the question referencing the contribution.
Part of the reason for putting this section here is that my SE profile is limited to 3000 characters, so I gave up on trying to squeeze it in there.

scikitguess entirely came about from my work with a paper by Jean Jacquelin. It was later supplemented by some Stack Overflow and Math Stack Exchange questions:
 Ndimensional spheres:
 https://stackoverflow.com/q/68302951/2988730 (JJ)
 https://stackoverflow.com/q/26574945/2988730
 https://stackoverflow.com/q/10344119/2988730
 https://stackoverflow.com/q/18934805/2988730
 https://math.stackexchange.com/q/820815/295281 (JJ)
 https://math.stackexchange.com/q/214661/295281
 https://math.stackexchange.com/q/2898295/295281 (JJ)
 Ndimensional ellipses:
 Ndimensional Gaussians:
 https://stackoverflow.com/q/25720600/2988730
 https://stackoverflow.com/q/27539933/2988730
 https://stackoverflow.com/q/27230824/2988730
 https://stackoverflow.com/q/67538941/2988730
 https://stackoverflow.com/q/8780912/2988730
 https://math.stackexchange.com/q/441422/295281 (JJ)
 https://math.stackexchange.com/q/1292889/295281 (JJ)
 Gaussian Mixtures:
 Exponentials:
 Sinusoids:
 https://stackoverflow.com/q/16716302/2988730
 https://stackoverflow.com/q/8126001/2988730
 https://stackoverflow.com/q/13405053/2988730
 https://math.stackexchange.com/q/36725/295281
 https://stackoverflow.com/q/42169643/2988730
 https://dsp.stackexchange.com/q/4886/26517
 https://stackoverflow.com/q/46296263/2988730
 https://stackoverflow.com/q/41191748/2988730
 https://stackoverflow.com/q/40221984/2988730
 https://stackoverflow.com/q/24086998/2988730
 https://stats.stackexchange.com/q/126273/102011
 Ndimensional spheres:

matplotlib.ticker.PrecentFormatter
added to matplotlib in PR #6251 in response to these posts: Yaxis: https://stackoverflow.com/a/36319915/2988730
 Xaxis: https://stackoverflow.com/a/36320013/2988730
 Not my answer, but still a good example after the fact: https://stackoverflow.com/q/51473993/2988730

Bug found in
matplotlib.transforms.AffineDeltaTransform
and fixed in https://github.com/matplotlib/matplotlib/pull/28375 Source: https://stackoverflow.com/q/78598087/2988730
 Added example to gallery: https://github.com/matplotlib/matplotlib/pull/28364

scipy.stats.iqr
added to scipy in PR #5805 partly just to write https://stackoverflow.com/a/38620147/2988730 
count
parameter added tonumpy.unpackbits
to make it invertible with respect tonumpy.packbits
in PR #10855 to improve this answer: https://stackoverflow.com/q/5602155/2988730 
Relaxed contiguity requirements for
numpy.ndarray.view
(see note about version 1.23.0) in PR #20722, which allows fornp.char.slice_
to exist in PR #20694: https://stackoverflow.com/q/70547027/2988730
 https://stackoverflow.com/q/53296394/2988730
 https://stackoverflow.com/q/39042214/2988730
 https://stackoverflow.com/q/40976714/2988730
 https://stackoverflow.com/q/64981711/2988730
 https://stackoverflow.com/q/31387047/2988730
 https://stackoverflow.com/q/69856133/2988730

haggis.numbers.english
implements the usage rules from this answer on English Language and Usage SE: https://english.stackexchange.com/a/111837/207127. 
haggis.math.full_width_half_max
implements a solution to https://stackoverflow.com/q/10582795/2988730 
My first attempt at a ufunc: is_integer_ufunc. This still needs work before it's ready for review by the numpy community, but the question crops up fairly frequently:

One of my first introductions to numpy was in the area of automatic histogram bin estimation in PR #, in which I added Doane's estimator: https://stats.stackexchange.com/q/55134/102011. Most of my research can be referenced to questions on Cross Validated:

This question led me to write
puzzlesolvers
: https://stackoverflow.com/q/56284249/2988730. Others that are nearduplicates: 
Added
haggis.math.segment_distance
as a result of the first of these questions: https://stackoverflow.com/a/70645020/2988730
 https://stackoverflow.com/q/39840030/2988730
 https://stackoverflow.com/q/23103345/2988730
With an interesting application here:

Angular distances are implemented in
haggis.math.ang_diff_pos
,haggis.math.ang_diff_min
andhaggis.math.ang_diff_abs
. They have appeared in 
haggis.npy_util.unmasked_index
, while not written specifically for https://stackoverflow.com/q/71389165/2988730, does provide an answer. Another possible application: https://stackoverflow.com/q/64082288/2988730

haggis.objects.getsizeof
was added to help with this question: https://stackoverflow.com/q/70792724/2988730, with the help of this one: https://stackoverflow.com/q/449560/2988730 
A question about formatting determinacy https://stackoverflow.com/q/41050445/2988730 led to a new test case for pyformat.info in PR #38.

haggis.objects.copy_class
is based on https://stackoverflow.com/q/49157255/2988730 
haggis.objects.copy_func
is based on https://stackoverflow.com/q/13503079/2988730 (duplicate of https://stackoverflow.com/q/6527633/2988730), implemented in https://stackoverflow.com/q/49076566/2988730 
haggis.load.load_module
is based on four answers: https://stackoverflow.com/a/43602645/2988730 (general loading technique)
 https://stackoverflow.com/a/67692/2988730 (general loading technique)

https://stackoverflow.com/a/50395128/2988730 (regisering in
sys.modules
)  https://stackoverflow.com/q/38647348/2988730 (variable injection)

haggis.logs.add_logging_level
is a productionized version of https://stackoverflow.com/a/35804945/2988730. 
haggis.files.docx.list_number
is based on the answers to https://stackoverflow.com/q/51829366/2988730. https://stackoverflow.com/q/77194722/2988730
 Mentioned by Steve Canny, author and maintainer of pythondocx, https://github.com/pythonopenxml/pythondocx/pull/582#issuecomment1743788250

haggis.args.SmartHelpFormatter
is a more flexible version of https://stackoverflow.com/a/76962505/2988730. 
haggis.recipes.chained_getter
mentioned in
Stuff that might get turned into PRs one day:
 Estimating the size of an index in numpy: https://stackoverflow.com/a/65857268/2988730

np.atleast_nd
function (numpy PR #18386):  Reverse winding of paths in matplotlib: https://stackoverflow.com/q/68275413/2988730
 Weighted partitioning in numpy: https://stackoverflow.com/q/20601872/2988730. This in turn will enable weighted histogram bin estimation.
 Weights to
np.std
: https://stackoverflow.com/q/2413522/2988730  Add a
count
method to Ufuncs returning booleans, likegreater_than
:  Justification for adding
np.same
(or special case ofcount
method onnp.equal
):  Wand's estimator for
np.histogram
: https://stats.stackexchange.com/a/114517/102011  Bin estimators for 2D histograms: https://stats.stackexchange.com/q/114490/102011
 ShimazakiShinomoto method of histogram bin estimation, possibly others: https://stats.stackexchange.com/q/798/102011
 Multiple lines to
plt.axvline
: https://stackoverflow.com/q/21300953/2988730  PGM reader/writer (for scipy.io?)
 Histogram of times (numpy filters down to pandas) might have been resolved by now:
 Speedup to
np.vander
: https://stackoverflow.com/q/48245987/2988730  Quaternion integrator: