​
System Architecture Overview

This document provides a high-level overview of the SG Cars Trends platform architecture, illustrating how the various components interact to deliver Singapore vehicle registration data and COE bidding results.

​
Architecture Diagram

​
System Components

​
Applications Layer

API Service (Hono Framework)
  • RESTful endpoints for data access (/v1/cars, /v1/coe, /v1/months)
  • tRPC integration for type-safe API calls
  • Workflow orchestration endpoints
  • Social media webhook handlers
  • Bearer token authentication
Web Application (Next.js 15)
  • User-facing interface with interactive charts and analytics
  • Blog functionality with LLM-generated content
  • Server-side rendering for optimal performance
  • Real-time analytics tracking
Documentation Site (Mintlify)
  • Comprehensive API documentation
  • Developer guides and examples
  • Interactive API exploration

​
Shared Packages

Database Package
  • Drizzle ORM schema definitions
  • Type-safe database queries
  • Migration management
  • Shared database client configuration
Types Package
  • Shared TypeScript interfaces
  • API request/response types
  • Domain model definitions
Utils Package
  • Common utility functions
  • Redis client configuration
  • Date formatting and calculations

​
Infrastructure Layer

AWS Services (Singapore Region - ap-southeast-1)
  • Lambda Functions: Serverless compute for API and web applications
  • CloudFront: Global content delivery network
  • API Gateway: RESTful API routing and management
  • RDS PostgreSQL: Primary data storage
  • Upstash Redis: Caching and session management
DNS Management (Cloudflare)
  • Domain routing and SSL termination
  • Global DNS resolution
  • DDoS protection and security features

​
Data Processing Workflows

QStash Orchestration
  • Scheduled data updates (hourly during business hours)
  • Workflow state management and error handling
  • Message queue processing
Cars Data Workflow
  • LTA DataMall integration for vehicle registration data
  • CSV processing and data transformation
  • Database updates with duplicate detection
COE Data Workflow
  • COE bidding results processing
  • Prevailing Quota Premium (PQP) data handling
  • Conditional blog generation when bidding cycles complete
Blog Generation Workflow
  • Google Gemini AI integration for content creation
  • Market analysis and insights generation
  • SEO optimization and metadata creation

​
External Integrations

Data Sources
  • LTA DataMall: Official Singapore vehicle and COE data
  • Google Gemini AI: LLM for automated blog content generation
Social Media Platforms
  • Discord: Webhook-based notifications
  • LinkedIn: API-based business updates
  • Telegram: Bot-based channel messaging
  • Twitter: API-based social media posts

​
Data Flow

  1. Data Ingestion: Scheduled workflows fetch data from LTA DataMall
  2. Data Processing: CSV files are parsed, validated, and transformed
  3. Data Storage: Processed data is stored in PostgreSQL with Redis caching
  4. Content Generation: AI generates blog posts from processed data
  5. Distribution: Updates are published to social media platforms
  6. User Access: Web application and API serve data to end users

​
Key Architectural Decisions

Monorepo Structure
  • Centralized codebase with shared packages
  • Consistent tooling and development experience
  • Type safety across application boundaries
Serverless Infrastructure
  • Cost-effective scaling based on usage
  • Minimal infrastructure management overhead
  • Regional deployment for low latency
Workflow-Based Processing
  • Reliable data processing with error recovery
  • Atomic operations with state management
  • Scheduled execution with manual override capability
Multi-Platform Social Integration
  • Diversified content distribution
  • Platform-specific content formatting
  • Graceful degradation with error notifications