ML-MLOps Portfolio Demo — 3 ML Services on GCP + AWS | Production Infrastructure

Production multi-cloud MLOps portfolio: 3 ML services deployed on GCP (GKE) and AWS (EKS).

🔗 Repository: https://github.com/DuqueOM/ML-MLOps-Portfolio
📄 Documentation: https://duqueom.github.io/ML-MLOps-Portfolio/
💼 LinkedIn: https://linkedin.com/in/DuqueOM

Stack: Python 3.11 · FastAPI · Docker · Kubernetes (GKE + EKS) · Terraform · GitHub Actions · MLflow · Prometheus · Grafana · SHAP · 395+ tests · 90-98% coverage

Timestamps:
0:00 - Multi-Cloud Infrastructure (GKE + EKS live)
0:11 - Live ML Predictions (SHAP, Sentiment, Demand)
0:46 - Experiment Tracking (MLflow)
1:02 - Observability & Load Testing (Grafana, Prometheus, Locust)
1:20 - CI/CD Pipeline & Code Quality
1:37 - Multi-Cloud Parity (GKE vs EKS side-by-side)
1:55 - Infrastructure as Code & DevSecOps
2:12 - Repository & GitHub Pages
2:35 - Summary & Contact

**Note (March 2026)**: This video shows the initial deployment. Since recording, I've implemented significant performance improvements:
- Async inference for BankChurn (ADR-015) — reduced errors from 81% to 0% under stress
- 50% CPU cost reduction (2000m → 1000m per pod)
- Full multi-cloud load testing (GCP + AWS)

See [load-test-results.md](docs/load-test-results.md) and [ADR-015](docs/decisions/015-async-inference-threadpool.md) for details.

Видео ML-MLOps Portfolio Demo — 3 ML Services on GCP + AWS | Production Infrastructure канала Duque Ortega Mutis