Worked on modularizing image processing within the ente-io/ente repository by introducing a dedicated image crate to isolate photo processing logic. The approach involved refactoring the existing architecture, renaming the main crate for improved clarity, and reorganizing modules into distinct areas such as machine learning and motion photo handling. Updated core dependencies by replacing libheic-rs with an Ente-owned heic-decoder, enhancing maintainability and ownership of the codebase. Leveraged Rust for both code refactoring and dependency management, laying a foundation for faster onboarding and scalable feature development. The work focused on architectural improvements rather than bug fixes, emphasizing long-term maintainability and clarity.