🌩️ Aether – AI-Powered Cloud Cost Optimization Platform

Aether is a cloud-native, AI-augmented platform designed to help platform teams identify and reduce overprovisioned infrastructure costs across Kubernetes and multi-cloud workloads.

---

🚀 Features

• 🤖 AI-based Resource Right-Sizing – Predicts optimal CPU/Memory for K8s workloads using Isolation Forest
• 📏 Rule-based Recommendations – Plain Python logic to suggest actions based on historical usage
• 📊 Interactive Dashboard – Streamlit UI to visualize savings opportunities per tenant
• 🔧 Terraform IaC Modules – EKS + Prometheus observability infrastructure
• 🔐 Multi-Tenant & GitOps Friendly – Secure, modular, and CI/CD ready

---

🧱 Project Structure

aether/
├── api/                   # FastAPI service (stubbed)
├── ai_engine/             # AI/ML logic (Isolation Forest model)
│   └── right_sizer.py
├── rules/                 # Pure Python rule engine
│   └── rule_engine.py
├── data_pipeline/         # (To be implemented: metric ingestion)
├── dashboard/             # Streamlit dashboard
│   └── app.py
├── infra/
│   └── terraform/         # IaC setup for EKS + Prometheus
│       ├── modules/
│       │   ├── eks/
│       │   └── prometheus/
│       └── environments/
│           └── dev/terraform.tfvars
├── pyproject.toml         # Poetry dependencies
├── docker-compose.yml     # Local dev setup
└── README.md              # You're here!

---

🧪 How to Run Everything

1. 🔧 Install Dependencies (Poetry)

poetry install

---

2. 🤖 Run the AI Right-Sizing Engine

poetry run python ai_engine/right_sizer.py

Outputs recommended CPU/memory for pods with confidence score based on usage data.

---

3. 📏 Run the Rule Engine

poetry run python rules/rule_engine.py

Evaluates sample workloads and prints rule-based recommendations like:

[RECOMMENDATION] analytics-pod: scale_down_cpu — CPU usage below 40% for 7+ days

---

4. 📊 Launch the Streamlit Dashboard

poetry run streamlit run dashboard/app.py

View recommendations in a tabular format with tenant-wise filtering.

---

5. 🛠️ Provision Infrastructure with Terraform

Setup

Update your terraform.tfvars with valid vpc_id and subnet_ids in:

aether/infra/terraform/environments/dev/terraform.tfvars

Run

cd aether/infra/terraform/modules/eks
terraform init
terraform apply -var-file=../../environments/dev/terraform.tfvars

cd ../prometheus
terraform init
terraform apply

---

🧠 Model Details

• Model: Isolation Forest
• Use: Detects underutilized workloads for downscaling
• Extension: Forecasting (Prophet, LSTM) planned in future versions

---

🔐 Security Principles

• Least-privilege IAM for AWS modules
• GitOps workflow compatible (ArgoCD/Flux ready)
• Future: Auth-enabled dashboards + multi-tenant K8s namespaces

---

📥 Contributing

We welcome:

• New ML-based optimization techniques
• Alerting & notification plugins
• GitHub PR-based automation logic

---

📄 License

MIT License — feel free to build on top of it.