Multi-service platform for turning any conversation, screenshot, or idea into an AI-assisted meme. The project currently consists of:

• web – Next.js 14 frontend with Tailwind CSS and NextAuth (Google OAuth)
• server – Go (Gin) API gateway orchestrating meme generation workflows
• services/ai – Python FastAPI stub that returns deterministic meme suggestions
• docker-compose.yml – Local development stack with Postgres and Redis foundations

• Node.js 20+
• Go 1.22+
• Python 3.11 (optional when running the AI service outside Docker)
• Docker Desktop (recommended for running the entire stack)

Copy each template before running locally:

cp web/.env.example web/.env.local
cp server/.env.example server/.env
cp services/ai/.env.example services/ai/.env

Populate the following values:

• NEXTAUTH_SECRET – random 32+ character string (openssl rand -base64 32)
• GOOGLE_CLIENT_ID / GOOGLE_CLIENT_SECRET – OAuth credentials from Google Cloud Console
• API_BASE_URL – typically http://localhost:8080
• DATABASE_URL – connection string for Postgres (docker-compose uses postgres://memify:memify@postgres:5432/memify?sslmode=disable)
• OPENAI_API_KEY – optional placeholder for future caption synthesis

docker-compose up --build

• Frontend – http://localhost:3000
• Go API – http://localhost:8080
• AI service – http://localhost:8000
• Postgres – localhost:5432 (user/password memify)
• Redis – localhost:6379

# Frontend
cd web
npm install
npm run dev

# Go API (requires Go installed locally)
cd server
go mod tidy   # generates go.sum
go run ./cmd/api

# AI service
cd services/ai
python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
uvicorn app.main:app --reload --host 0.0.0.0 --port 8000

• web/src/app/workspace – authenticated meme generation workspace hitting /api/memes/generate
• server/internal/httpserver – Gin router, middleware, handlers
• server/internal/ai – HTTP client for the Python AI service
• services/ai/app – FastAPI endpoints and Pydantic schemas mirroring the Go contracts

1. Persist user sessions, context snippets, and meme generations using Postgres + migrations (e.g., Atlas, Goose, or Prisma schema driving sqlc).
2. Implement screenshot OCR and humor classification inside the AI service; connect to a template embedding store (pgvector or Redis vector).
3. Render meme images through a dedicated rendering worker (Go image libraries or Node canvas) and store assets in S3-compatible storage.
4. Expand frontend workspace with template chooser, caption editor, and history dashboard.
5. Add background queue processing via Redis (Asynq) for heavier pipelines and notifications.