• 

  Aries Agent

• 

  Aries Architecture

• 

  Aries Overview

TCC RAPTOR ARIES — AI-Powered Retail Virtual Try-On Platform

Smarter shopping for customers, actionable intelligence for sellers — powered by the ARIES multi-agent AI system on Google Gemini.

🔗 Live demo: https://tcc-raptor-retail-ru2czwsr6a-el.a.run.app

Created by Deepan Raj — Senior Solution Architect (Azure, AWS & GCP) · TCC RAPTOR · deepanrey@gmail.com

---

What is this?

An AI retail platform: customers browse, search in natural language, and try clothing on their own photo (Gemini 3.1 Flash Lite Image), while sellers get an AI dashboard with forecasts, dynamic pricing, and a copilot. The ARIES multi-agent orchestrator coordinates the search, recommendation, try-on, outfit-builder, notification, and seller-copilot agents.

Highlights: natural-language product search · personalized recommendations & discounts · AI virtual try-on · outfit builder · seller intelligence dashboard · multi-language UI · Aries voice chatbot.

---

🚀 How to Run

Prerequisites

• Node.js 18+ and npm
• A Google Gemini API key (for the Virtual Try-On feature) — create one in Google AI Studio

1. Install dependencies

npm install

2. Configure environment

Create a .env file in the project root:

VITE_GEMINI_API_KEY=your_gemini_api_key

VITE_* variables are compiled into the app at build time. .env is git-ignored — never commit it. In production the key is injected from Google Secret Manager (see Deployment).

3. Start the dev server

npm run dev

Then open the printed URL (Vite defaults to http://localhost:5173).

4. Build for production (optional, local)

npm run build     # outputs static site to dist/
npm run preview   # serve the production build locally

All scripts

Command	What it does
npm run dev	Start the Vite dev server with hot reload
npm run build	Production build → dist/
npm run preview	Preview the production build locally
npm run typecheck	Type-check with TypeScript
npm run lint	Lint with ESLint

---

🧩 Architecture

The frontend is a React SPA; the browser calls Gemini and Firebase Auth directly, while the app is hosted on Google Cloud Run. The ARIES orchestrator coordinates the AI agents.

📐 Full architecture → ARCHITECTURE.md

Deploy pipeline & runtime request flow, build-time secret handling, and the IAM model — with detailed diagrams.

---

☁️ Deployment

Containerized (multi-stage Docker → Nginx) and running on Google Cloud Run. Every push to main auto-builds and deploys through Cloud Build, with the Gemini key injected from Secret Manager at build time.

git push origin main   # → Cloud Build → new Cloud Run revision (~4 min)

🚀 Full deploy runbook → DEPLOYMENT.md

Step-by-step: GCP project setup, Secret Manager, Artifact Registry, the Cloud Build trigger, and Cloud Run.

---

🛠️ Tech Stack

Layer	Technology
Frontend	React 18 · TypeScript · Vite · Tailwind CSS · React Router · Lucide
AI	Google Gemini 3.1 Flash Lite Image
Backend	Firebase (Firestore · Auth · FCM · Storage)
Hosting	Google Cloud Run (Nginx, scale-to-zero)
CI/CD	GitHub → Cloud Build → Artifact Registry → Cloud Run
Secrets	Google Secret Manager

---

📁 Project Structure

.
├── src/
│   ├── components/    # UI components (Header, modals, ProductCard, VirtualTryOn…)
│   ├── pages/         # Routes (Home, Product, Category, seller/*)
│   ├── context/       # Auth, Cart, SellerAuth React contexts
│   ├── lib/           # aiSearch, firebase, ariesEngine
│   ├── data/          # Product catalog & seller mock data
│   └── public/        # Static assets (product images + README diagrams)
├── Dockerfile         # Multi-stage build → Nginx (port 8080)
├── nginx.conf         # Static serving + SPA fallback
├── cloudbuild.yaml    # CI/CD pipeline: build → push → deploy
├── ARCHITECTURE.md    # As-built architecture & pipeline (diagrams)
└── DEPLOYMENT.md      # Step-by-step deploy runbook

---

📚 Documentation

Document	What's inside
ARCHITECTURE.md	How the deploy pipeline & runtime work — diagrams, secret handling, IAM
DEPLOYMENT.md	Step-by-step deploy runbook (GCP → Secret Manager → Cloud Build → Cloud Run)

---

License

Proprietary to TCC RAPTOR. All rights reserved.

Built With

• cloudrun
• firebase
• firestore
• gemini3.1
• github
• google-cloud
• googlescretmanager
• javascript
• react
• tailwindcss
• typescript
• vite