optimize

To identify performance bottlenecks and optimize slow code, you need profiling-driven hotspot detection to find issues like nested loops, N+1 queries, and unnecessary allocations, then apply targeted improvements that preserve existing behavior.

The best way to reduce latency and improve throughput is by applying risk-aware optimizations with documented trade-offs, such as sacrificing memory for lower latency, and validating the changes with tests and before/after benchmarks.

You can optimize concurrency and memory inefficiency without breaking functionality by performing behavior-preserving validation with tests, ensuring that targeted improvements like blocking synchronous I/O removal maintain existing application behavior.

You need a deep performance and reliability review for code refactoring when profiling reveals complex throughput, latency, memory, or concurrency bottlenecks that require a targeted deep dive to safely apply high-impact optimizations.

Code optimization does not strictly require measuring before and after performance, but applying simple before/after benchmarks is highly recommended to validate expected results and confirm the targeted improvements yield safe gains.