The continuous evolution of malicious software presents a growing challenge for cybersecurity defenses. While traditional signature-based detection struggles against obfuscation and novel threats, machine learning offers promising capabilities for identifying malware through automated static analysis. This seminar focuses on how static properties of executables—such as byte-level patterns, control-flow structures, imported APIs, and features derived from disassembled code—can be leveraged for effective detection. We will discuss and build various machine learning approaches, including feature-based, byte-level, and end-to-end detectors, and examine their respective trade-offs in terms of accuracy, robustness, and interpretability.