Week 10 Worklog

Week 10 Objectives:

  • Perform a major UI/UX overhaul using Tailwind CSS and Ant Design to modernize the application’s interface.
  • Refactor the AI OCR processing pipeline for high compatibility with AWS Fargate Serverless (CPU-only migration).
  • Enhance invoice validation transparency with the BusinessValidationSummary component and version history tracking.

Tasks to be carried out this week:

DayTaskStart DateCompletion DateReference Material
2Project - UI/UX Overhaul & Auth Flows:
• Execute major UI/UX updates across the entire application
• Redesign and implement Landing Page, Login, and Register flows
• Integrate Tailwind CSS and Ant Design for modern styling		16/03/202616/03/2026
3Project - Architecture Refinement:
• Refine backend architecture for SmartInvoice Shield
• Optimize resource allocation for AWS services		17/03/202617/03/2026
4Project - Business Validation & UX:
• Build BusinessValidationSummary component for detailed error display
• Merge “Check” and “Risk” tabs into a unified “Validation Results” tab
• Integrate Version History and warning banners for replaced invoices
• Hide raw ErrorCodes and fix JSON casing discrepancies in frontend		18/03/202618/03/2026
5Project - AWS Infra Optimization:
• Finalize cost-optimized Event-Driven Architecture
• Integrate Amazon SQS for asynchronous OCR and VietQR processing
• Configure SSL/TLS (ACM), Route 53, and CloudWatch Alarms		19/03/202619/03/2026
6Project - AI Container Refactoring (Fargate):
• Resolve libcublas.so missing library errors in container
• Migrate OCR Dockerfile from GPU (8GB) to CPU-only base image (1.5GB)
• Switch to CPU-only paddlepaddle/torch for 100% Fargate compatibility		20/03/202620/03/2026
7Project - OCR Parsing & Validation Sync:
• Fix corrupted stringified dictionaries in Python OCR output
• Resolve early-exit bug in InvoiceProcessorService.cs to ensure full reporting
• Fix “Total amount 0 VND” display error and add missing XML warnings		21/03/202621/03/2026

Week 10 Achievements:

  • Dramatically reduced AI container size (from 8GB to 1.5GB) ensuring efficient Fargate deployment.
  • Successfully integrated asynchronous SQS processing for OCR and VietQR flows to resolve performance bottlenecks.
  • Improved data accuracy by resolving OCR parsing discrepancies and refining backend validation logic.