Dynostic

Problem Statement

How might we make the classification of label placement and quality inspection data tracking as accurate as possible with the help of artificially intelligent software?

Solution

1. Mobile application to help inspectors manage multiple QC entries at once

2. Web page for managers to view analytics, as well as change system controls

3. AI inspection service detects whether labels pass or fail the quality control process

Challenges:

1. The dataset was quite small which made it harder to create a highly accurate model

2. Real-time Data Handling: Efficiently managing real-time data communication and processing, particularly with WebSockets and cloud functions.

3.Scalability: Balancing the load and scaling the system as usage grows, especially in a cloud environment.

4. User Experience: Designing an intuitive and responsive interface for both the mobile inspector app and the analytics dashboard.

Key Decisions:

1. AI Inspection: Utilizing Cloud Vision for multi-label image detection and binary classification, optimizing for speed with a two-step verification process (First Check and Second Check) to ensure label accuracy and placement.

2. Authentication Service: A standalone service to manage user logins, ensuring secure and isolated access across multiple applications.

3. App Server: Centralizing business logic, enabling CRUD operations, and supporting real-time communication using WebSockets with Rails ActionCable, showcasing a focus on efficient data management and user interaction.

4. RESTful API Design: Adhering to REST principles such as client-server decoupling, statelessness, and uniform interfaces, indicating a commitment to scalable and maintainable web services.

5. Google Cloud Integration: Strategic use of Cloud Run, Cloud SQL, Pub/Sub, Cloud Functions, and BigQuery, demonstrates a deep integration with Google Cloud services, ensuring scalability, reliability, and efficient data handling.