For most Python library authors, rewriting performance-critical code in Rust has historically meant a steep learning curve, fragile tooling, and uncertain payoff. That calculus has shifted. This talk argues that 2026 represents a genuine tipping point, driven by the convergence of three forces: PyO3 and maturin have made Python↔Rust bindings straightforward to set up and ship; the Apache Arrow ecosystem provides a shared memory model between Python and Rust that eliminates costly serialization overhead; and LLMs have dramatically reduced the cost of writing idiomatic Rust for developers who aren't fluent in the language.
The talk is structured around a practical framework for library authors. We start with the question of fit: what characteristics make a library a strong candidate for a Rust rewrite — and what makes it a poor one. We then walk through the end-to-end process as it stands today, from project setup with maturin to CI, packaging, and distribution on PyPI. Finally, we look honestly at performance: where Rust delivers dramatic gains, and where the overhead of crossing the Python↔Rust boundary erodes them. Two open-source libraries are used as concrete illustrations throughout.
Attendees will leave with a clear-eyed framework for evaluating whether a Rust rewrite makes sense for their own libraries, and a realistic picture of what the journey involves in 2026. No prior Rust experience is assumed.