AI‑Powered Predictive Autoscaling & Cost Optimization for Cloud Infrastructure. AutoScalerAI forecasts demand, makes proactive scaling decisions, and automates infrastructure changes to keep latency low and cloud bills under control.
- Live Demo:
https://ramiadell.github.io/ai-with-devops/ - Status: Frontend landing page (no backend) – ideal for portfolio and concept showcase
AutoScalerAI is a DevOps + AI concept that predicts traffic using time‑series models and scales your infrastructure before the spike hits. It integrates with familiar tooling (Terraform, Kubernetes, Prometheus) and can either automate changes or propose them for human review.
- Predictive Autoscaling: Forecasts traffic using models like Prophet/LSTM to pre‑scale services.
- Intelligent Cost Optimization: Rightsizes capacity, blends spot/on‑demand, minimizes idle waste.
- Proactive Resource Allocation: Pre‑warms services/caches and shifts load ahead of hotspots.
- Automated Provisioning & Scaling: Applies infra updates via Terraform/Ansible, updates HPAs/ASGs.
- Deep Integrations: Terraform • Kubernetes • Prometheus • Grafana • ArgoCD • OpenAI • Python.
Pipeline stages:
- Data Ingestion → Metrics from Prometheus, CloudWatch, logs, custom events.
- AI Prediction Engine → Time‑series forecasting (Prophet/LSTM) on resource usage and demand.
- Decision Layer → Policy & risk budgets determine safe, cost‑efficient actions.
- Automation Layer → Terraform/Ansible, HPAs/ASGs, cloud APIs apply changes.
- Visualization → Grafana dashboards, Slack bot notifications and summaries.
The landing page includes an inline SVG diagram and section summaries for each stage.
- Collect metrics (Prometheus, CloudWatch)
- Predict demand with AI (Prophet, LSTM)
- Trigger scaling decisions (policy + SLO budgets)
- Automate infrastructure changes (Terraform, Ansible / K8s HPA)
- Visualize insights (Grafana, Slack Bot)
- Reduced Downtime: Pre‑scale to avoid SLO violations during bursts.
- Lower Cloud Costs: Scale down intelligently during lulls; rightsize continuously.
- Smarter Infrastructure: Policy‑driven decisions anchored in SLOs and risk tolerance.
- Automated Operations: Less toil; more time for product and reliability work.
- Infra: Terraform, Ansible, Kubernetes (HPA/VPA)
- Observability: Prometheus, Grafana
- CI/CD: ArgoCD (GitOps)
- AI: Python models (Prophet, LSTM), optional OpenAI for reasoning/assistive decisions
- Messaging: Slack bot for change proposals and summaries
This repository hosts a static, responsive landing page demonstrating the concept. It uses Tailwind CSS via CDN and minimal JavaScript for smooth scrolling and scroll‑triggered animations.
Just open index.html in your browser. No build step required.
This site is deployed as a GitHub Pages project site.
- Commit
index.htmlat repo root. - GitHub → Settings → Pages → Build and deployment → Source:
Deploy from a branch. - Select
mainand folder/root→ Save. - It will be available at
https://<username>.github.io/<repo>/.
Deployed example: https://ramiadell.github.io/ai-with-devops/
- Update CTA button links in the "Contact" section of
index.html(GitHub URL, email). - Edit content in each section to reflect your environment and preferences.
- Tailwind is loaded via CDN; no build pipeline required.
- High‑contrast dark theme with a DevOps aesthetic.
- Smooth scrolling and IntersectionObserver‑based reveals (respects
prefers-reduced-motion). - Keyboard focus states preserved for interactive elements.
- “An AI‑powered autoscaler that predicts traffic and proactively scales infra, reducing downtime and cost.”
- “Integrates Prometheus, Terraform, Kubernetes, and a forecasting model to make scaling intelligent rather than reactive.”
- “Balances cost and reliability using policy‑driven decisions and risk budgets.”
Ideas, feedback, and improvements are welcome. If you’d like to collaborate or turn this into a working prototype, open an issue or reach out.
- GitHub: add your repository link in the CTA of
index.html. - Email: replace
[email protected]in the CTA with your address.