Skip to main content

​
Architecture Overview

Trendteller is built as a comprehensive multi-part system that aggregates e-commerce data from multiple platforms, transforms it through a data pipeline, and provides insights through a modern web interface and AI-powered analytics.

​
System Components

Platform Frontend

Nuxt 3 dashboard for analytics and insights

Data Pipeline

Dataform transformations on BigQuery

Data Connectors

Airbyte integrations for 9 e-commerce platforms

AI Workflows

Kestra scripts for insights and forecasting

​
Data Flow Diagram

┌─────────────────────────────────────────────────────────────────┐
│                     E-Commerce Platforms                        │
│  (Bling, VNDA, Shoppub, Tiny, Microvix, Braavo, JetERP, etc.) │
└────────────────┬────────────────────────────────────────────────┘
                 │
                 ▼
┌─────────────────────────────────────────────────────────────────┐
│              Airbyte (Custom Connectors)                        │
│          15 Sources + 2 Destinations (TypeScript)               │
└────────────────┬────────────────────────────────────────────────┘
                 │
                 ▼
┌─────────────────────────────────────────────────────────────────┐
│                 BigQuery Data Lake                              │
│                  (Bronze Layer - Raw Data)                      │
└────────────────┬────────────────────────────────────────────────┘
                 │
                 ▼
┌─────────────────────────────────────────────────────────────────┐
│                    Dataform Transformations                     │
│    Bronze → Silver (Standardized) → Gold (Analytics-Ready)     │
└────────────────┬────────────────────────────────────────────────┘
                 │
                 ▼
┌─────────────────────────────────────────────────────────────────┐
│                         Hasura V2                               │
│       GraphQL API Engine (Postgres/NeonDB + BigQuery)          │
└────────────────┬────────────────────────────────────────────────┘
                 │
                 ▼
┌─────────────────────────────────────────────────────────────────┐
│                  Platform Front (Nuxt 3)                        │
│          Analytics Dashboard + Forecasting UI                   │
│                    (Auth0 + Netlify)                            │
└─────────────────────────────────────────────────────────────────┘

              ┌──────────────────────┐
              │  Kestra Scripts      │
              │  (AI Insights,       │
              │   Forecasting,       │
              │   Web Crawling)      │
              └──────────────────────┘

​
Data Flow Process

1

Data Ingestion

Airbyte connectors extract data from 9 e-commerce platforms (11 brands) including Bling, VNDA, Shoppub, Tiny, Microvix, and others.
2

Raw Storage

Raw data lands in BigQuery’s Bronze layer, maintaining source format and structure for full traceability.
3

Data Transformation

Dataform processes data through Silver (standardized) and Gold (analytics-ready) layers using SQLX templates.
4

API Exposure

Hasura V2 exposes transformed data via a type-safe GraphQL API, combining BigQuery analytics data with Postgres operational data.
5

Data Consumption

Frontend dashboard displays analytics and visualizations. AI scripts generate insights, forecasts, and automated reports.

​
Technology Stack

​
Frontend Stack

  • Framework: Nuxt 3 (v3.16.2) with TypeScript
  • UI Library: Vuetify 3 component library
  • State Management: Pinia with persistence
  • Authentication: Auth0
  • API Client: GraphQL with Genql (type-safe)
  • Hosting: Netlify

​
Data Stack

  • Data Warehouse: Google BigQuery
  • Transformation: Dataform Core 3.0.0
  • Integration: Airbyte CDK (TypeScript)
  • Orchestration: Kestra workflows
  • API Layer: Hasura V2 GraphQL Engine

​
AI & Analytics

  • AI Providers: OpenAI, Google Gemini AI
  • Automation: Puppeteer for web crawling
  • Database: PostgreSQL (NeonDB) for operational data

​
Infrastructure

  • Google Cloud Platform: BigQuery data warehouse
  • AWS: Additional cloud services
  • Neon: Managed PostgreSQL database

​
Key Features

Multi-Brand Support

Consolidates data from 11 brands across 9 e-commerce platforms into unified analytics

Real-time GraphQL API

Type-safe API powered by Hasura V2 with automatic schema generation

AI-Powered Insights

Leverages OpenAI and Gemini AI for forecasting and trend analysis

Scalable Data Pipeline

Medallion architecture on BigQuery ensures data quality and performance

​
Scale & Metrics

  • Brands: 11 supported brands
  • Integrations: 9 e-commerce platforms
  • Connectors: 15 source + 2 destination connectors
  • Frontend Pages: 23 pages
  • UI Components: 29 reusable components
  • Data Tables: 20+ analytics tables

​
Next Steps

Key Concepts

Learn about medallion architecture and multi-brand analytics

API Reference

Explore the Trendteller API

Architecture Deep Dive

Explore detailed system architecture

API Reference

Integrate with Trendteller’s GraphQL API