PyDev of the Week: Logan Thomas

This week we welcome Logan Thomas as our PyDev of the Week! If you’d like to see some of the things that Logan has been up to, you can do so on Logan’s GitHub profile or by visiting his personal website. Logan works at Enthought as a teacher. If you haven’t done so, you should check out their list of courses.

Now let’s spend some time getting to know Logan better!

Can you tell us a little about yourself (hobbies, education, etc):

I currently work for Enthought as a Scientific Software Developer and Technical Trainer. Previously, I worked for a protective design firm as a machine learning engineer and I’ve also been employed as a data scientist in the digital media industry.

My educational background consists of mathematics, statistics, and mechanical engineering. During my final semester of graduate school at the University of Florida, I took a data mining class from the Department of Information Science and was hooked. I’ve worked in data science ever since.

I’m passionate about software and love to (humbly) share my knowledge with others. Working for Enthought gives me the opportunity to help students, and larger organizations, accelerate their journey toward greater digital maturity. We focus on digital transformation – transforming an organization’s business model to take full advantage of the possibilities that new technologies can provide. It’s so fulfilling being able to help scientists and engineers develop the digital skills and tools they need to be able to automate away the trivial tasks and free their minds of this cognitive load to focus on the things they love – the science and engineering behind the problems they face.

Outside of work, I love to stay active by running, brewing a good cup of coffee, and have recently started smoking brisket / pork (a must if you live in Texas like me).

Why did you start using Python?

I was originally hesitant to get into software development as both my parents are software engineers and I wanted my own career path. But, with their encouragement, I took an Introduction to Programming course during my undergraduate degree. The class was taught in Visual Basic and given it was my first true taste of any type of software development, I didn’t have the best experience. Thus, I decided to stay away from computer science and ended up going to graduate school for mechanical engineering. I’m glad the story didn’t end there.

In one of my mechanical design courses, we used a CAD (computer-aided design) program that supported Python scripting (if you knew how to use it!). I remember spending an entire weekend running simulations by hand with Microsoft Excel. When I asked my friend how he managed to get his testing done so quickly, he showed me roughly 10 lines of Python code he wrote to automate the simulation loop. After that exchange, I made it my goal to give computer science another try and take advantage of the superpower a general-purpose programming language like Python could provide.

What other programming languages do you know and which is your favorite?

Since graduate school, I’ve worked in the data science space for about 8 years – all of which my primary programming language has been Python. I love Python for its ease of use and ability to stretch across domain areas like scripting / automation, data analysis / machine learning, and even web design. For me, Python is the clear favorite.

Besides Python, I have the most experience with MATLAB, R, SQL, and Spark. I’ve dabbled in a few other languages as well: Scala, Ruby, C++, HTML / CSS.

What projects are you working on now?

Currently, I am working on building a curriculum for Enthought Academy – a scientific Python catalog of courses for R&D professionals. We have some really great courses in the queue for 2023! I’m putting final touches on our new Data Analysis course that covers Python libraries like NumPy, pandas, Xarray, and Awkward Array.

Personally, I’m working on a new project called Coffee & Code. The idea is to share small code snippets and video shorts for exploring the Python programming language — all during a morning coffee break. My first major project will be trying to teach my brother, who has 0 programming experience, the Python programming language. I plan to film it live and release videos as we walk through installation, environment setup, and language syntax together.

I’m also on the planning committee for two major Python conferences: PyTexas and SciPy Conference. I love both of these events and can’t wait for their 2023 appearance.

Which Python libraries are your favorite (core or 3rd party)?

There are so many to choose from! From the standard library, I have to give a shout-out to the collections module and itertools. The namedtuple, double-ended queue (deque), Counter, and defaultdict are gamechangers. The itertool recipes in the official Python documentation are a goldmine.

My favorite third-party libraries from the scientific Python stack are NumPy and scikit-learn. I use PyTorch and Tensorflow with Keras (in that order) for deep learning. I also have to give credit to IPython. It is my default choice for any type of interactive computing and is quite honestly my development environment of choice (when partnered with Vim). One lesser-known library that I have come to enjoy is DEAP (Distributed Evolutionary Algorithms in Python). I worked with this library extensively for a genetic programming project in the past.

I see you are an instructor at Enthought. What sorts of classes do you teach?

We teach all sorts of classes at Enthought. Given that all of our instructors started out in industry first, we can boast that our curriculum is “for scientists and engineers by scientists and engineers”. We have introduction to Python and programming courses as well as more advanced machine learning and deep learning courses. We also have a software engineering course, data analysis course, data management course, and desktop application development course. All of these live in what we call a learning path or “track”. We have an Enthought Academy Curriculum Map that displays the various tracks we offer: Data Analysis, Machine Learning, Tool Maker, and Manager.

What are some common issues that you see new students struggling with?

I love Python but can admit that there are some rough edges when first getting started. Most students struggle with understanding virtual environments and why they need one if they’ve never encountered dependency management before. Other students are surprised at how many third-party libraries exist in the ecosystem. I often get questions on which ones to use, how to know which ones are safe, and where to look for them.

Some students struggle with IDEs and which development environment to use. I’ve gotten questions like “which is the single best tool: JupyterLab, IPython, VSCode, PyCharm, Spyder, …”. For new students, I try to help them understand that these are tools and there may not be a single correct answer. Sometimes I reach for VSCode when needing to navigate a large code base. If I’m sharing an analysis with a colleague or onboarding a new analyst, I may reach for a Jupyter Notebook. If I’m prototyping a new algorithm, I’ll probably stay in IPython. But, these are my preferences and may not be the same for all students. When looking at these as tools in a toolbelt, rather than a single permanent choice, it becomes easier to pick the right one for the task at hand.