The discovery and synthesis of various proteins have significantly contributed to scientific advancement by deepening our understanding of biological processes and enabling medical breakthroughs. However, certain proteins that pose risks to human health require early identification and classification to mitigate potential hazards and guide targeted research. Tools like AlphaFold have revolutionized protein structure prediction, achieving remarkable accuracy and offering valuable insights into protein functions. Despite its groundbreaking capabilities, AlphaFold’s reliance on significant computing power and large-scale protein datasets limits its usability in high-throughput classification tasks and situations demanding rapid analysis. To address these limitations, we propose a lightweight machine learning model tailored for six key protein classes—HYDROLASE, TRANSFERASE, OXIDOREDUCTASE, VIRAL PROTEINS, IMMUNE SYSTEM-RELATED PROTEINS, and TRANSCRIPTION FACTORS—each playing a critical role in biological processes such as metabolism, immune response, and gene regulation. Our model aims to balance efficiency and accuracy in protein classification while minimizing computational complexity, making it suitable for broader and resource-constrained applications. Future work will validate the model using larger, more diverse datasets, integrate structural features, and incorporate advanced deep learning techniques to enhance prediction accuracy and expand its applicability, including in protein-drug interaction research.