MSc Dissertation Research In Progress

ML-Based Anomaly Detection in DevSecOps Pipelines

Building a system that detects security threats that rule-based approaches miss by applying machine learning to real-time log analysis.

Tech Stack

PythonIsolation ForestGraylogOpenSearch

Key Features

  • Real-time anomaly detection with 30-second detection windows
  • 10 attack scenarios designed to evaluate ML detection vs rule-based approaches
  • Measuring precision, recall, and false positive rates
  • Focus on reducing false positives while maintaining high threat detection

Directly applicable to AI safety monitoring—the same techniques that detect anomalous pipeline behavior can detect anomalous model behavior.

Infrastructure as Code Complete

Secure Auto-Scaling AWS Infrastructure

Production-ready AWS infrastructure using modular Terraform—VPC, load balancing, auto-scaling, and monitoring, all as code.

Tech Stack

TerraformAWS (VPC, EC2, ALB, ASG, S3, IAM, CloudWatch, SNS)Nginx

Key Features

  • Multi-AZ VPC with public/private subnets
  • Application Load Balancer distributing traffic
  • Auto Scaling Group with Launch Templates
  • CloudWatch alarms + SNS notifications
  • Modular code structure (vpc, ec2, alb, asg, monitoring modules)
  • Stress testing to validate auto-scaling behavior

Demonstrates security-first infrastructure design with proper network segmentation, least-privilege IAM, and comprehensive monitoring.

More Projects Coming

Currently working on additional projects in AI security and ML pipeline protection. Check back soon or follow my writing for updates.