Sharon Roy developed an ML-driven smart substitution dataset expansion for the DiscountMate_new repository, focusing on building robust data foundations to enhance substitution recommendations. She engineered a large, detailed product catalog dataset using CSV, emphasizing data engineering and dataset management to ensure scalability and machine learning readiness. Sharon integrated comprehensive documentation and dataset artifacts, streamlining onboarding and supporting future model development. Her work established the groundwork for improved substitution accuracy by delivering ML-ready data artifacts across multiple sprints. Although the project was limited to one feature over a month, the depth of dataset expansion and documentation provided a solid base for future ML initiatives.