From Science to Open Source and Back Again
summary
A practiced path between lab and open source shows concrete results: the Bluesky Project and stewardship of matplotlib produce reusable stacks rather than one-off scripts; a tracking library processes terabytes of high‑speed imagery at 3000 frames per second to follow ~20,000 particles per frame; that library appears across domains ~300× and facility tooling supports ~550 papers per year from NSLS-II. Emphasis is on actual code, design choices (library vs application), measurable reuse, and whether the system runs reliably in production.
AI-written summary · about
chapters (20)
- 0:00Introduction
- 1:08Open source mentors
- 2:12Software in modern science
- 3:35Manual to automated analysis
- 5:46Automating frost drop imaging
- 7:51Dynamics from high-speed imagery
- 8:52Collaborative software impact
- 9:57Myth of lone genius
- 11:19Matplotlib's paper impact
- 13:06Library vs application mindset
- 17:57Library vs application code
- 20:25Library mindset guide
- 23:14Co-designing data systems
- 25:22Open source maintainer role
- 27:51Enabling science indirectly
- 28:54Research software engineers
- 31:21Version control importance
- 32:04Advice for scientist devs
- 33:07Guiding new developers
- 34:11Q&A
related talks
52mKeynote: Learning programming and science with Scientific Python
01h 18mKeynote
45mKeynote: Stories of Science, Community, Failures and Successes
30mEmpowered by Python - A success story
44mKeynote
29mResearch Software Stories: Four Years of Lessons in Sustainable Science
01h 16mReproducible machine learning and science with python
30mOn the Fly and On the Flight: Scientific Data Analysis Beyond the Beamline