02 — Selected projects
Things I built
to prove the point
Detection engineering, BI, and machine learning — all public on GitHub, all reproducible.
Detection engineering
Microsoft Sentinel · KQL · MITRE ATT&CK
Microsoft Sentinel SOC Home Lab
Three production-style analytics rules for the attacks a Tier-1 analyst triages every shift:
brute-force authentication (T1110), impossible-travel sign-ins (T1078) and after-hours privileged
activity (T1078.004 / T1098). Each detection ships the KQL, the tuning notes, an ATT&CK mapping
and a step-by-step investigation runbook — the same artifact a real SOC keeps for every alert.
View on GitHub →
Business intelligence
Power BI · DAX
Sales, Marketing & HR Dashboards
Three interactive Power BI dashboards with slicers, drilldowns and clean DAX measures — YoY revenue,
CAC, marketing ROI and attrition rate — over reproducible datasets.
View on GitHub →
Machine learning
Python · scikit-learn
Loan Approval Prediction
Logistic regression, decision trees and random forest compared on a loan-approval dataset, with the
full preprocessing chain and evaluation metrics kept in the repo rather than in a notebook comment.
View on GitHub →
Data engineering
Python · pandas · Matplotlib
Sales & Customer ETL Pipeline
Extract, transform and load over sales and customer data, ending in aggregated revenue by region —
the unglamorous layer every dashboard quietly depends on.
View on GitHub →
Classification
Python · scikit-learn
Breast Cancer Classifier
Logistic regression on the breast-cancer dataset, reported honestly: confusion matrix, ROC curve and
the metrics that matter when false negatives are expensive.
View on GitHub →
Unsupervised
Python · K-Means · PCA
Wine Clustering Analysis
K-Means over the Wine dataset with PCA for dimensionality reduction, plus the visualisations that
make the cluster assignments arguable rather than asserted.
View on GitHub →
Tooling
Python · Tkinter · SQLite
Employee Management System
A desktop CRUD application — Tkinter front end, SQLite behind it. Built for the same reason SOC
automation gets built: someone was doing it by hand.
View on GitHub →
Scraping
Python · BeautifulSoup
IMDb Ratings Scraper
Pulls titles, ratings and genres from IMDb with Requests and BeautifulSoup, then charts the ratings
distribution. Small, but it is the acquisition step of every analysis above.
View on GitHub →