Cybersecurity Network Intrusion Detection Using Machine Learning

This project targeted brute-force network intrusions and focused on detecting malicious FTP and SSH activity from the CSE-CIC-IDS2018 dataset.

The work combined model comparison, feature engineering, and deployment thinking rather than stopping at a single benchmark score.

• Evaluated Logistic Regression, Random Forest, SVM, and a Keras/TensorFlow deep neural network against the same dataset and validation workflow.
• Performed extensive exploratory analysis, preprocessing, scaling, feature selection, and class-balance handling across 1M+ records.
• Containerized the final workflow with Docker to keep the pipeline reproducible and easier to deploy.

This project broadened the ML work beyond NLP into security analytics while reinforcing data-heavy evaluation and deployment discipline.