While the growth of online marketplaces has significantly increased both the volume and variety of products available, finding items that fit users' individual preferences has become increasingly challenging. This paper proposes an AI-based Hybrid E-Commerce Product Recommendation System that adopts the concept of integrating rule-based with machine learning-driven similarity measures to offer personalized, context-aware product recommendations. The proposed system makes use of four different metrics for similarity: category similarity, price similarity, tag similarity, and textual similarity. These are computed using the TF-IDF vectorization method and cosine similarity to capture semantic relationships between product descriptions. These metrics will be aggregated using a weighted hybridization strategy to provide the best trade-off in terms of accuracy and diversity for recommendations.