Improving Dashboard Refresh and Resolving AI Service Integration Issues

 During this development cycle, I focused on improving the PharmaStock dashboard experience and troubleshooting several critical issues with the AI prediction service.

On the frontend, I implemented both automatic and manual refresh functionality for the dashboard. The auto-refresh uses an interval set to 60 seconds, ensuring that inventory statistics and AI predictions stay up to date without requiring user interaction. In addition, I added a manual refresh option that triggers the same centralized refresh logic. This logic consolidates all dashboard updates into a single function, making it easier to maintain and extend.

While integrating AI predictions into the dashboard, I encountered several issues with the AI service. Initially, the environment was using Python 3.14, which caused dependency installation failures, specifically with pydantic-core. The error required Rust for compilation, which was not intended for this setup. I resolved this by installing Python 3.11, recreating the virtual environment, and properly configuring the project. I also ensured that the .venv directory was added to .gitignore to avoid unnecessary version control issues.

After resolving the environment setup, I ran into additional problems with backend communication. The prediction endpoints for reorder alerts and expiration risks were returning 500 errors or hanging indefinitely. The backend logs showed that requests were being sent to the AI service, but no responses were received. This resulted in TaskCanceledException errors due to HttpClient timeouts.

These issues highlighted a breakdown in communication between the backend and the AI service. Although the AI service appeared to be running, it was not properly responding to requests. This required further debugging of request handling and service responsiveness to ensure reliable integration.

Overall, this week involved both feature development and deep debugging. The dashboard refresh improvements enhanced the user experience, while resolving environment and service communication issues laid the groundwork for stable AI integration moving forward.


Comments

Post a Comment

Popular posts from this blog

PharmaStock: Backend Infrastructure & Database Connectivity

 This week marked the transition from planning and design into real backend development. After organizing the project structure and defining use cases in previous weeks, I focused on building the technical foundation that the rest of the application will depend on. The primary goal was completing SCRUM-16, which involved connecting the ASP.NET Core backend API to a MySQL database and verifying that the system could successfully generate and manage database tables. I began by configuring MySQL to run inside a Docker container. Using Docker ensures that every developer on the team works against the same database environment, which reduces setup inconsistencies and avoids the common “it works on my machine” problem. Once the container was running correctly, I verified that it could be started, stopped, and inspected reliably. Next, I integrated Entity Framework Core with the Pomelo MySQL provider and registered the DbContext within the application. This step allows the backend to m...
Read more

PharmaStock Backend Progress Complete all backend endpoints

Image
 I completed the full Medication API (GET, POST, PUT, PATCH, DELETE) and finished SCRUM-37, implementing the stock quantity adjustment endpoint. Inventory quantities can now be increased or decreased, negative values are prevented, and all changes persist correctly in the database. I also finalized the inventory creation flow with duplicate lot validation and verified everything through Swagger testing. On the frontend side, I started SCRUM-38 and began building the Angular inventory list component to connect the UI to the new backend functionality. Overall, this week solidified the backend foundation and began transitioning into full-stack development. Medication Post Workflow            Medication Post          Post Inventory for new medication               Inventory adjustments
Read more

PharmaStock: Sprint Planning Complete

This week was focused on finalizing project planning and getting everything ready for development. I completed the setup of the Jira board , finalized all sprints for the full length of the project, and added clear completion criteria for each sprint and task. This gives the team a clear roadmap from MVP through enhancement features. I also synced my local GitHub setup with the main project repository and verified that the branching structure is ready for feature development. With planning complete and source control aligned, the project is now fully prepared to move into active development next week.
Read more