2025-10-21 –, UVM Alumni House Silver Pavilon
45 min talk going through the decision tree for picking Python environment tools / package managers geared toward Scientific computing and context on why Python environment management is so difficult -- spoiler: it's why Python is so popular -- for its flexibility and extensibility.
What if every Python environment you created just worked—on any platform, six months later, with GPU dependencies handled automatically? This talk tackles the reproducibility crisis facing the Python data science community and provides the decision-making frameworks practitioners desperately need.
We'll cut through the confusion of pip vs Poetry vs conda vs pixi vs uv with a practical guide based on real-world constraints: Do you have binary dependencies? Need cross-platform support? GPU requirements? Regulatory compliance needs?
You'll walk away with a decision tree you can apply immediately, migration guides for your current projects, and confidence that your environments will actually reproduce when you need them to.
