perf-optimization

Reduce memory allocations in hot paths by applying pattern-based fixes like pre-allocation, pooling, and batching. These reusable optimization patterns target allocation reduction directly from heap trace evidence to lower garbage collection overhead.

Prioritize performance fixes by ranking candidate code changes using an impact-to-risk ratio. This approach evaluates profiling data to decide when to execute high-impact changes and when to defer or skip risky micro-optimizations entirely.

Benchmark code changes before and after optimization by implementing one fix at a time with tests-first, then running comparisons using the exact same fixed configuration parameters and stop-when heuristics to verify measurable latency improvements.

Performance profiling handles runtime noise and benchmark artifacts by applying specific interpretation rules for cumulative versus flat profiles and artifact partitioning. This isolates true production cost from misleading benchmark environmental data.

Avoid caching or library swaps for hot path optimization when explicit safety guardrails are violated, particularly in security-sensitive contexts. The workflow includes anti-pattern prevention to ensure optimizations do not introduce security vulnerabilities.