Initially, MIDAS approached us after a year of falling behind on projects with a cheaper vendor. Recognizing the critical nature of time for a startup, we swiftly delivered on front-end projects, creating user interfaces for new machines used in hospitals and medical facilities for drug waste collection and diversion detection. These projects utilized technologies such as Angular, Docker, Git, Java, Jira, NodeJS, and PostgreSQL. We also implemented performance optimization techniques like lazy loading and code splitting, and developed new Micro Frontend components, adhering to principles for independent deployment and scaling. Our successful delivery earned their trust, leading to a request for a Proof of Concept (POC) for AI-powered video analysis. We used object detection and frame processing for narrative generation in detecting diversion from videos. We leveraged Python, Conda, Poetry, YOLO, AWS Bedrock, and Claude to achieve high-confidence assessments of theft occurrence. The success of this POC expanded our engagement, including the development of a new AI diversion detection product, data science initiatives, and more.