SpendLens

AWS cost aggregation and recommendation engine for multi-team organizations.

Ingests AWS billing CSVs (monthly or on-demand), aggregates costs by service/team/environment, and generates 4 types of cost optimization recommendations: idle resource detection, storage class optimization, instance right-sizing, and reserved instance candidates.

Evidence

CSV Ingestion & Parsing — Python batch worker parses both CUR (Cost and Usage Report) and simplified billing CSV formats. Supports AWS standard columns (date, service, region) plus custom tags (team, project, environment, resource_id).

(worker/csv_parser.py)

Cost Aggregation — Daily and monthly cost rollups grouped by:

• Service (EC2, S3, RDS, Lambda, etc.)
• Team (via tag)
• Environment (production, staging, dev, via tag)
• Region (optional)

Computed aggregations stored in Supabase PostgreSQL.

(worker/aggregator.py)

Recommendation Engine — Analyzes aggregated spend and generates 4 recommendation types:

1. Idle Resources — Services with zero usage hours over lookback window
2. Storage Optimization — S3 storage classes + EBS volume consolidation
3. Right-Sizing — High-cost services with low utilization (compute instances, RDS)
4. Reserved Instances — Steady workloads eligible for RI purchases (cost/benefit analysis)

Each recommendation includes: estimated monthly savings, implementation effort, confidence score.

(worker/recommendation_engine.py)

Multi-Tenant Isolation — Supabase PostgreSQL with Row-Level Security (RLS) policies. Each user sees only their organization's costs and recommendations.

(supabase/migrations/ — RLS policies on costs, recommendations, budgets tables)

Budget Alerts — Set per-service/per-team/per-environment cost thresholds. Alerts via email (Resend integration) when budget exceeded.

(supabase/functions/send-budget-alert.ts)

Dashboard — React 18 + Recharts visualization:

• Daily/monthly cost trends by service
• Service-level cost breakdown
• Top-cost services ranking
• Budget consumption progress
• Recommendation list with implementation guidance

(src/pages/DashboardPage.tsx, src/components/CostTrendChart.tsx)

What-If Simulator — Model cost changes from configuration updates (storage class changes, instance downsizing, retention policy adjustments). Projects estimated savings.

(src/pages/SimulatorPage.tsx)

Stack — React 18, Vite, Supabase (PostgreSQL + Auth + Edge Functions), Python 3.9+ (pandas, psycopg2), Tailwind CSS, shadcn-ui, Recharts.

How It Works

1. Export AWS billing CSV → Download from AWS Cost Management or use Cost and Usage Reports (CUR)
2. Load into SpendLens → Run Python worker: python3 batch_aggregator.py billing.csv --org-id ORG_ID
3. Worker processes → Parser normalizes CSV → Aggregator computes daily/monthly rollups → Recommendation engine generates suggestions
4. Data persists to Supabase → All team members see aggregated costs + recommendations via dashboard
5. User sets budgets → Define per-service or per-team thresholds → Alerts trigger if exceeded
6. Optimize → Review recommendations → Use simulator to model changes → Update cloud infra

Getting Started

Prerequisites

• Supabase account (PostgreSQL database)
• AWS billing CSV (export from AWS Cost Management or CUR)
• Python 3.9+ (local worker processing)

Frontend Setup

npm install
npm run dev  # http://localhost:5173

Backend Setup

cd worker
pip install -r requirements.txt
cp .env.example .env
# Edit .env with Supabase connection string

Load Sample Data

# Generate 30 days of synthetic billing data
python3 scripts/generate_sample_billing.py --days 30 --output billing.csv

# Process with worker
python3 batch_aggregator.py billing.csv --org-id demo-org

# Open dashboard: http://localhost:5173

Custom CSV Processing

python3 batch_aggregator.py /path/to/your/billing.csv --org-id your-org-id

CSV Format

Parser supports two formats:

Simplified (recommended for testing):

date,service,cost,usage,unit,region,team,project,environment,resource_id
2024-01-15,EC2,125.50,100,Hours,us-east-1,backend,web-app,production,i-123456
2024-01-15,S3,45.20,500,GB-Month,us-west-2,data,analytics,production,

AWS CUR (standard format):

• Automatically detected and normalized by parser
• Supports all CUR columns + custom tags

Testing

# Frontend
npm run test

# Backend
cd worker
python3 -m pytest tests/

Deployment

Frontend

npm run build
# Deploy to Vercel, Netlify, or any static host

Worker

• GitHub Actions — Schedule CSV processing via Actions workflow
• Cron — 0 2 * * * python3 batch_aggregator.py /billing/latest.csv --org-id ORG_ID
• Docker — Build and deploy worker as container with scheduled triggers

Architecture

AWS Billing CSV
      ↓
Python Worker Pipeline
  ├─ csv_parser.py   → Normalize CSV
  ├─ aggregator.py   → Daily/monthly rollups
  └─ recommendation_engine.py → Generate 4 recommendation types
      ↓
Supabase PostgreSQL (RLS-protected)
      ↓
React Dashboard
  ├─ Cost trends (Recharts)
  ├─ Service breakdown
  ├─ Budget alerts
  └─ What-if simulator

Security

• Row-Level Security — All tables (costs, recommendations, budgets) protected by RLS policies
• Multi-tenant isolation — Users see only their organization's data
• No secrets in git — .env file gitignored, credentials in Supabase Secrets Manager

License

MIT