Python Performance Optimization Patterns

About

This skill provides a set of battle-tested Python performance patterns designed to accelerate data-intensive applications with minimal risk of breaking existing logic. It guides Claude to implement high-impact optimizations such as LRU caching for file metadata, NumPy vectorization for overlap matrices, O(n) top-K selection with argpartition, and multi-threaded image loading. By focusing on verified speedups from real-world performance audits, this skill helps developers identify and fix common bottlenecks in I/O-bound and numerical computing tasks while providing clear guidance on when to avoid certain optimizations to maintain accuracy.

Key Features

• Efficient directory traversal using single-scan iterdir patterns
• Vectorization of nested loops using specialized NumPy built-ins like np.add.at
• 0 GitHub stars
• Intelligent metadata caching to reduce redundant filesystem calls
• Implementation of thread-safe parallel I/O for batch image loading
• Optimized top-K selection via O(n) complexity argpartition

Use Cases

• Speeding up batch file operations in directories with high file counts
• Accelerating scientific image processing pipelines that handle large datasets
• Optimizing data-heavy NumPy calculations involving pixel-wise overlap counting