Revenue Leakage v2 — Railway + Supabase Deployment Guide

Deploy the feat/kpi-pipeline-feature-flags-synthetic branch from
https://github.com/Techtorch/revenue-leakage-v2

---

1. Supabase (Database)

1. Go to supabase.com → New Project
2. Pick a region close to your Railway region (e.g. US East)
3. Set a strong database password — you'll need it below
4. Once created, go to Settings → Database → Connection string → URI
5. Copy the connection string — it looks like:

   postgresql://postgres.[project-ref]:[PASSWORD]@aws-0-us-east-1.pooler.supabase.com:6543/postgres
   

   Use the Transaction (port 6543) pooler URI for the backend.

---

2. Railway — Backend Service

1. Go to railway.app → New Project → Deploy from GitHub repo

2. Select Techtorch/revenue-leakage-v2, branch feat/kpi-pipeline-feature-flags-synthetic

3. Settings:

   • Root Directory: rev-leak-agents
   • Builder: Dockerfile (auto-detected from rev-leak-agents/Dockerfile)
   • Start Command: leave blank (Dockerfile CMD handles it)

4. Environment Variables:

   Variable	Value
   RL_DB_DSN	postgresql://postgres.[ref]:[PASSWORD]@aws-0-us-east-1.pooler.supabase.com:6543/postgres
   RL_RUN_MIGRATIONS_ON_STARTUP	true
   RL_OPENAI_API_KEY	sk-... (your OpenAI key for GPT-4o extraction)
   RL_CORS_ORIGINS	https://<frontend-domain>.up.railway.app (set after frontend deploys)
   RL_LOG_LEVEL	info
   PORT	8001

5. Networking: Generate a public domain (e.g. rev-leak-backend-production.up.railway.app)

6. Deploy — first deploy runs Alembic migrations automatically

---

3. Railway — Frontend Service

1. In the same Railway project → New Service → Deploy from GitHub repo

2. Select same repo + branch

3. Settings:

   • Root Directory: revenue-leakage-v2/remix-of-revenue-leakage-ui
   • Builder: Dockerfile (auto-detected)

4. Environment Variables:

   Variable	Value
   VITE_API_BASE_URL	https://rev-leak-backend-production.up.railway.app

   This gets baked in at build time. If you change the backend URL, redeploy the frontend.

5. Networking: Generate a public domain (e.g. rev-leak-ui-production.up.railway.app)

6. Deploy

---

4. Post-Deploy Checklist

• Update backend RL_CORS_ORIGINS with the actual frontend Railway URL
• Hit https://<backend>/health — should return {"status": "ok"}
• Open frontend URL — dashboard should load (empty until you run pipeline)
• Click "Load CITI Data" on Ingestion page to process CITI documents
• Click "Process Synthetic" to run synthetic test documents through KPI pipeline
• Verify Dashboard shows KPI signals, leakage categories, and trend chart
• Verify Knowledge Graph shows document relationships (not all orphans)

---

5. Optional: Basic Auth

To protect the API with HTTP Basic Auth, set these on the backend service:

Variable	Value
RL_BASIC_AUTH_USER	your-username
RL_BASIC_AUTH_PASS	your-password

The /health endpoint is excluded from auth (for Railway healthchecks).

---

Architecture

┌─────────────┐     HTTPS      ┌─────────────────┐     SQL      ┌──────────┐
│  React/Vite │ ──────────────→ │  FastAPI Backend │ ───────────→ │ Supabase │
│  (Railway)  │  VITE_API_URL   │    (Railway)     │  RL_DB_DSN   │ Postgres │
└─────────────┘                 └─────────────────┘              └──────────┘
                                        │
                                        │ RL_OPENAI_API_KEY
                                        ▼
                                 ┌──────────────┐
                                 │  OpenAI API  │
                                 │  (GPT-4o)    │
                                 └──────────────┘

---

Troubleshooting

• Migrations fail: Check RL_DB_DSN is correct and Supabase allows connections from Railway IPs
• CORS errors: Ensure RL_CORS_ORIGINS matches the exact frontend URL (including https://)
• Empty dashboard: Run the pipeline first via Ingestion page buttons
• OpenAI errors: Verify RL_OPENAI_API_KEY is set and has GPT-4o access