Coca-Cola SmartLabel

System Overview

• Event-driven pipeline ingests CSV product data from S3 and transforms it into structured workloads
• Queue-based processing distributes screenshot generation tasks across Lambda executions
• Headless browser automation captures SmartLabel pages and stores images in S3
• DynamoDB persists revision history and brand metadata for query-based retrieval
• API Gateway exposes archive data through HTTP endpoints
• React single-page application provides a guided interface for searching and browsing archived records
• Frontend deployed as static assets to S3, fully decoupled from backend compute

Archiver (Backend System)

• Serverless ingestion and archival pipeline built on AWS managed services
• Parser Lambda converts CSV files into structured records and partitions workloads
• SQS queue decouples ingestion from processing, enabling asynchronous execution
• Archiver Lambda uses Puppeteer with headless Chromium to capture product page screenshots
• S3 stores both source files and generated screenshot artifacts
• DynamoDB models revision history and brand relationships based on query patterns
• API Gateway provides access to archived data for downstream consumers

Browser (Frontend System)

• React single-page application for searching and exploring archived SmartLabel data
• Guided filter workflow supports date range, brand, sub-brand, and GTIN queries
• Dynamic retrieval of brand and taxonomy data from backend APIs
• Paginated visual results with screenshot previews and metadata display
• Modal-based inspection enables detailed review without disrupting search flow
• Client-side JSON export supports lightweight data extraction
• Statically deployed to S3 for low-cost, scalable frontend delivery

Architectural Decisions

• Adopted event-driven architecture to decouple ingestion, processing, and consumption layers
• Introduced SQS-based workload partitioning to handle Lambda execution constraints
• Separated backend pipeline from frontend application via API-driven integration
• Modeled DynamoDB schema around access patterns to support efficient filtering and retrieval
• Used serverless browser automation to capture visual archive evidence at scale
• Deployed frontend as static assets to align with API-first cloud architecture

The platform was designed to address the need for reliable archival and retrieval of SmartLabel product data over time.

Key challenges included:

• Processing large datasets within serverless execution limits
• Capturing visual snapshots of product pages at scale
• Designing a system that supports both archival storage and user-driven retrieval
• Integrating multiple AWS services into a cohesive, maintainable architecture
• Providing a usable interface for exploring archived data

Outcome

• Delivered a complete cloud-native platform spanning ingestion, processing, storage, and user interaction
• Transformed a complex archival workflow into a scalable, asynchronous system
• Enabled both programmatic and user-facing access to historical SmartLabel data
• Demonstrated end-to-end ownership across backend architecture, frontend development, and cloud infrastructure