Software Reverse Engineering

Software reverse engineering infers a program’s logic, interfaces, and security properties from compiled code, especially stripped or obfuscated binaries, supporting maintenance, malware analysis, and vulnerability discovery. Methods combine disassembly and static analysis with dynamic execution, emulation, fuzzing, and source-to-binary matching; neural decompilers and large language model agents increasingly assist with pseudocode recovery and analyst workflows. Evaluation spans function-matching and decompilation benchmarks, challenge binaries, and executable tests, since readability, compilation, and similarity scores do not guarantee behavioral equivalence. Work increasingly uses tool-using agents and execution-guided repair, but obfuscation, scale, trustworthy ground truth, prompt-injection risks, and the gap between model and expert performance remain central challenges.