uv now handles the author’s package installs, virtual environments, lock files, Python versions and command runs in one workflow, replacing pip, virtualenv and Poetry.
10-100x speed gains versus pip—based on Astral benchmarks—plus automatic .venv creation, pyproject.toml management and uv.lock generation drove the switch to the Rust-based tool.
Commands such as uv init, uv add and uv run cut out manual environment activation, while uv sync and uv python pin 3.12 keep dependencies and interpreter versions aligned.
Older projects do not need a full migration because uv pip install -r requirements.txt lets developers adopt uv gradually before moving new projects to its full workflow.
The author still recommends conda for many data-science beginners and says pip remains worth learning, but argues uv is the cleaner default for new Python projects.