CompletedAWSPythonSageMaker+3 more
Cloud Anomaly Detection
Scalable ML-powered cloud monitoring solution detecting anomalies in real-time resource usage using the Isolation Forest algorithm, AWS SageMaker, Lambda, and SNS alerts.
Timeline
2024
Role
Cloud / ML Engineer
Team
Solo
Status
CompletedTechnology Stack
AWS
Python
SageMaker
Lambda
SNS
Machine Learning
Key Challenges
- Tuning Isolation Forest for noisy resource metrics
- Keeping inference serverless and cheap with Lambda
- Alerting without false-positive fatigue
Key Learnings
- Unsupervised anomaly detection in production
- AWS SageMaker training and deployment
- Event-driven architectures with Lambda and SNS
Cloud Anomaly Detection
Overview
A scalable, ML-powered cloud monitoring solution that detects anomalies in real-time resource usage using the Isolation Forest algorithm. Models are trained and hosted with AWS SageMaker, inference runs through Lambda, and anomalies trigger automated SNS alerts.
Features
- Real-time anomaly detection on cloud resource metrics
- Isolation Forest — unsupervised, robust to unlabeled data
- AWS SageMaker for model training and hosting
- Serverless inference with Lambda
- Automated alerting via SNS