Engineering: Platform Architecture & Tech Stack #12

New issue

Closed

opened 2026-05-02 18:52:49 +00:00 by agent-cos · 2 comments

agent-cos commented

2026-05-02 18:52:49 +00:00

Objective

Design the technology platform architecture for WardrobeAI. This covers the customer-facing app, the vendor portal, the curation engine backend, and the operational tools.

System Components

1. Customer App (Web + Mobile)

Style interview flow
Box preview + approval
Feedback system (per item, per box)
Profile management (invariants, budget, frequency)
Order history + wardrobe tracker
Return initiation
Referral program dashboard
Payment management

2. Vendor Portal (Web)

Application/onboarding
Inventory catalog management (upload products, sizing, photos)
Order notifications + fulfillment
Return tracking
Analytics dashboard (what's selling, customer feedback, fit data)
Payment/invoicing

3. Curation Engine (Backend Service)

Style profile matching algorithm
Budget optimization
Inventory selection + outfit coherence
AI design-on-demand generation
Constraint/invariant enforcement
Learning model (per-customer + cross-customer)

4. Operations Dashboard (Internal)

Box assembly queue
Vendor order management
Return processing
Customer support tools
Inventory management
Financial reporting

5. Analytics & Data Pipeline

Customer behavior tracking
Keep/return rate analytics
Vendor performance metrics
Style trend analysis
Financial KPIs

Tech Stack (Recommended)

Frontend

Web: Next.js (React) — SEO-friendly, fast, TypeScript
Mobile: React Native (shared codebase with web) or PWA initially
UI: Tailwind CSS + Radix UI (accessible, fast to build)

Backend

API: Node.js with Fastify or Python with FastAPI
Database: PostgreSQL (relational data) + Redis (caching/sessions)
Auth: NextAuth or Clerk (quick startup setup)
Payments: Stripe (subscription billing + one-time charges)
File Storage: S3-compatible (Storj or AWS S3 for product images)

AI/ML

Curation: Python service with scikit-learn → evolve to PyTorch as data grows
Style matching: Vector embeddings (pgvector in PostgreSQL or Pinecone)
Design-on-demand: Image generation API (Midjourney, DALL-E, or fine-tuned SDXL)
NLP for invariants: LLM API (OpenAI, Anthropic, or local model) for parsing free-form constraints

Infrastructure

Hosting: Vercel (frontend) + Railway or Fly.io (backend) — low cost startup
CI/CD: GitHub Actions → Forgejo Actions (once runner is solid)
Monitoring: Umami (already deployed) + Sentry for errors
Email: Resend or Postmark
SMS: Twilio (for shipping notifications)

MVP Scope (First 3 months)

Customer web app: style interview + feedback only (no mobile app yet)
Vendor portal: simple inventory upload + order notifications
Curation engine: rule-based (no ML yet)
Operations: manual box assembly + spreadsheet tracking
No AI design-on-demand in MVP (Phase 2)

Deliverable

Architecture doc at engineering/platform-architecture.md — system diagram, API design, database schema, tech stack decisions with rationale, MVP scope definition, and phased rollout plan.

## Objective Design the technology platform architecture for WardrobeAI. This covers the customer-facing app, the vendor portal, the curation engine backend, and the operational tools. ## System Components ### 1. Customer App (Web + Mobile) - Style interview flow - Box preview + approval - Feedback system (per item, per box) - Profile management (invariants, budget, frequency) - Order history + wardrobe tracker - Return initiation - Referral program dashboard - Payment management ### 2. Vendor Portal (Web) - Application/onboarding - Inventory catalog management (upload products, sizing, photos) - Order notifications + fulfillment - Return tracking - Analytics dashboard (what's selling, customer feedback, fit data) - Payment/invoicing ### 3. Curation Engine (Backend Service) - Style profile matching algorithm - Budget optimization - Inventory selection + outfit coherence - AI design-on-demand generation - Constraint/invariant enforcement - Learning model (per-customer + cross-customer) ### 4. Operations Dashboard (Internal) - Box assembly queue - Vendor order management - Return processing - Customer support tools - Inventory management - Financial reporting ### 5. Analytics & Data Pipeline - Customer behavior tracking - Keep/return rate analytics - Vendor performance metrics - Style trend analysis - Financial KPIs ## Tech Stack (Recommended) ### Frontend - **Web**: Next.js (React) — SEO-friendly, fast, TypeScript - **Mobile**: React Native (shared codebase with web) or PWA initially - **UI**: Tailwind CSS + Radix UI (accessible, fast to build) ### Backend - **API**: Node.js with Fastify or Python with FastAPI - **Database**: PostgreSQL (relational data) + Redis (caching/sessions) - **Auth**: NextAuth or Clerk (quick startup setup) - **Payments**: Stripe (subscription billing + one-time charges) - **File Storage**: S3-compatible (Storj or AWS S3 for product images) ### AI/ML - **Curation**: Python service with scikit-learn → evolve to PyTorch as data grows - **Style matching**: Vector embeddings (pgvector in PostgreSQL or Pinecone) - **Design-on-demand**: Image generation API (Midjourney, DALL-E, or fine-tuned SDXL) - **NLP for invariants**: LLM API (OpenAI, Anthropic, or local model) for parsing free-form constraints ### Infrastructure - **Hosting**: Vercel (frontend) + Railway or Fly.io (backend) — low cost startup - **CI/CD**: GitHub Actions → Forgejo Actions (once runner is solid) - **Monitoring**: Umami (already deployed) + Sentry for errors - **Email**: Resend or Postmark - **SMS**: Twilio (for shipping notifications) ## MVP Scope (First 3 months) - Customer web app: style interview + feedback only (no mobile app yet) - Vendor portal: simple inventory upload + order notifications - Curation engine: rule-based (no ML yet) - Operations: manual box assembly + spreadsheet tracking - No AI design-on-demand in MVP (Phase 2) ## Deliverable Architecture doc at `engineering/platform-architecture.md` — system diagram, API design, database schema, tech stack decisions with rationale, MVP scope definition, and phased rollout plan.

agent-cos referenced this issue

2026-05-02 18:52:49 +00:00

Business: Phased Roadmap (Concept → Launch → Scale) #14

agent-cos referenced this issue

2026-05-02 19:35:17 +00:00

Project Management: Milestone-Based Issue Organization #28

sscoble commented

2026-05-25 02:12:45 +00:00

Owner

✅ Design/planning ticket with substantial content. Contains full system architecture with 5 components (Customer App, Vendor Portal, Curation Engine, Ops Dashboard, Analytics), detailed tech stack recommendations per layer, and MVP scope definition. Keeping open as a feature.

✅ **Design/planning ticket with substantial content.** Contains full system architecture with 5 components (Customer App, Vendor Portal, Curation Engine, Ops Dashboard, Analytics), detailed tech stack recommendations per layer, and MVP scope definition. Keeping open as a feature.

agent-engineering commented

2026-05-25 14:03:49 +00:00

✅ COMPLETED: Platform Architecture document written at engineering/platform-architecture.md (11.5KB). Covers: system architecture diagram, 5 system components (Customer App, Vendor Portal, Curation Engine, Operations Dashboard, Analytics Pipeline), full tech stack with alternatives, database schema (customers, boxes, box_items, vendors, inventory), API design with endpoints for all services, MVP scope (first 3 months), Phase 2-3 roadmap.

✅ COMPLETED: Platform Architecture document written at `engineering/platform-architecture.md` (11.5KB). Covers: system architecture diagram, 5 system components (Customer App, Vendor Portal, Curation Engine, Operations Dashboard, Analytics Pipeline), full tech stack with alternatives, database schema (customers, boxes, box_items, vendors, inventory), API design with endpoints for all services, MVP scope (first 3 months), Phase 2-3 roadmap.