Скачать или смотреть Predictive Autoscaling on Kubernetes with LSTM | Live Demo & Monitoring

This video showcases the project I completed during my Master’s internship in Computer Systems Engineering at Akli Mohand Oulhadj University (Bouira, Algeria).

Provisioned a Kubernetes cluster (Ubuntu 20.04) with Docker, Calico, MetalLB, and shared NFS storage.

Designed & built a resource‑consumption prediction system for Kubernetes micro‑services using an LSTM Encoder–Decoder model.

Generated training data by applying controlled CPU load and streaming metrics (via top) into InfluxDB.

Implemented LSTM‑driven predictive autoscaling—dynamically adjusting both vertical resources and horizontal replicas.

Deployed & configured Grafana dashboards for real‑time visualisation of predictions.

Containerised and deployed every module—data generation, training, inference, visualisation, and autoscaling—on the cluster.
***-------------------------------------------------------***
For full source code and technical details, see my GitHub repository: **github.com/AdelBouzidi**.
***-------------------------------------------------------***