AI FinOps Platform is an AI-powered platform for cloud cost optimization and forecasting. Built with FastAPI, Python, and modern MLOps tools, it allows teams to track multi-cloud usage (AWS, Azure, GPC), detect anomalies, and predict future expenses using real-time data and machine learning.
Key features:
- Unified ingestion pipelines for each provider
- Powerful REST API with FastAPI & OpenAPI docs
- Interactive React dashboard (Next.js + TailwindCSS)
- ML-driven forecasting & anomaly detection
- Full IaC deployment via Terraform & Helm
- Terraform (infrastructure provisioning)
- Kubernetes + Helm (orchestration)
- Docker (containerization)
- GitHub Actions (CI/CD)
- Prometheus + Grafana (monitoring)
- Cloud Billing APIs (AWS, Azure, GCP)
- Python (FastAPI)
- PostgreSQL (relational DB)
- InfluxDB (time series data)
- Forecasting: XGBoost, statsmodels
- Clustering: KMeans
- Anomaly Detection: Isolation Forest, Autoencoders
- Recommendation: Reinforcement Learning models
- React + TailwindCSS (Next.js)
- Data visualization with Recharts / Chart.js
- Optimize cloud resource usage and cost efficiency
- Predict monthly spending using machine learning
- Detect anomalous cost spikes and resource misusage
- Provide actionable AI-based cost-saving recommendations
- Offer a user-friendly dashboard for Finance & Tech teams
- Enhanced API docs with detailed parameter descriptions & examples
- Ingestion: Added per-provider scripts, unified CSV loader and header-only fallback
- Makefile: New targets for
ingest-api,fetch-aws,fetch-azure,fetch-gcp - Infrastructure: Updated Terraform modules partially.
- Security: Enforced CORS policies and OAuth2/JWT authentication on backend
- User Manual: Expanded with Docker Compose, Jupyter, and CLI workflows
- Added
providerfield to all cost endpoints and UI filters - Switched to stacked bar & multi-series line charts for richer insights
- Introduced infinite scroll and deduplication in cost tables
- Proxying
/docs&/redocthrough Next.js for consolidated UX - Updated complete Helm modules for deployment in AWS.
# 1. Clone repo
git clone https://github.com/Impesud/ai-finops-platform.git
cd ai-finops-platform
# 2. Setup venv & install
make init
# 3. Ingest sample data
make fetch-aws
make fetch-azure
make fetch-gcp
# or via API
make ingest-api
# 4. Run app
make dev
# 5. Explore
- Dashboard: http://localhost:3000
- Swagger: http://localhost:3000/docs
- ReDoc: http://localhost:3000/redoc
# 6. Deploy Pipeline on AWS (Helm + EKS)
# To provision the full environment
# (EKS cluster, IAM, ALB controller, Helm setup, etc.)
make full-setup
# To deploy or update the platform on an existing cluster
# (after building & pushing images)
make deploy app/ # FastAPI backend
frontend/ # Next.js dashboard
services/ # ETL & ingestion modules
scripts/ # Ingestion & deployment scripts
notebooks/ # ML notebooks
infra/ # Terraform & Helm
app/data/ # Generated CSVs
docs/ # Documentation & images
tests/ # Pytest suites
Makefile # Task automation
README.md # Project overviewMaintainer: Erick Jara — CTO & AI/Data Engineer
📧 erick.jara@hotmail.it | 🌐 GitHub: Impesud