MLflow Dashboard Installed to Monitor AI Health, Immediately Spammed with 'User Typed Banana' Alerts

In the final part of a three-part series on building production AI, the focus shifts to deployment and monitoring. The AI system, designed to generate Nigerian Adire-style art, is launched with a user-friendly interface built using Gradio, allowing users to input prompts and receive generated images. The system is integrated with MLflow, enabling performance data to be tracked and monitored. The model is deployed on HuggingFace, a platform for AI models, and a Model Card is created to provide instructions and quality scores. A GitHub Action is used to automate testing, evaluation, and deployment, ensuring the model meets quality standards. A Monitoring Dashboard is also built to track performance and alert developers if generation time exceeds 30 seconds. By leveraging techniques such as torch.compile and xFormers, the system achieves a speedup ratio of up to 2.0, reducing GPU time and costs. The completed system is now production-ready and available for use.