Computers that smell and Panel apps without a server

Many cool developments happening across the Python world!

Hello fellow datanistas!

Having survived two crazy weeks of one new daycare bug per week, I'm finally over the hump. So, I am pleased to share October's Data Science Programming Newsletter edition!

Making Computers Smell

This is work by a friend of mine, Alex Wiltschko, who has been doing incredibly creative computational work since before I met him. His latest creation, done with his teammates at Google Brain, is figuring out how to map the odours of small molecules computationally.

Dhingy by Ned Batchelder

Boston Python User Group's fearless leader, Ned Batchelder, has put out a tool for summarizing GitHub activity called Dhingy. You can install and run it locally. If you're feeling creative, add it to a repo with a GitHub Action that automatically generates digests daily on cron.

Discovering Algorithms with AlphaTensor

Who would have thought that the humble matrix multiply could have been improved? Not me. DeepMind's new creation, AlphaTensor, figures out how to make matrix multiplies more efficient.

Panel Apps in the browser without a server

Python in the browser continues its year! This time with Panel, the dashboarding package from Anaconda, can now be compiled down to a serverless WebAssembly static (dynamic?) site thanks to PyScript and pyodide. The ability to deploy a Panel app without a server is a massive win for Python app development. I think this is also the first high-power use of PyScript (at least that I’ve seen); kudos to Phillip Rudiger & colleagues for making it happen!

Career skills: both so-called "hard" and "soft" skills

This Twitter thread by Sorin Sbârnea clarifies how technical and people skills are necessary for a data scientist's career progression. But it turns out that technical skills are needed early on for mostly credibility. On the other hand, what makes senior people effective are the people skills that help move projects forward and deliver actionable results, which ultimately matter. So my takeaway is that if you're in a junior role right now, level up your technical skillsets so that you have credibility points to spend on people relationships later.