Explore real-world data science & machine learning work. Each post breaks down the approach and takeaways — and how the same methods can uncover opportunities, reduce risk, and guide better decisions.
Unsupervised ML flags suspicious behavior (USB events, file access) to surface potential insider risks early. This Project uses unsupervised anomaly detection (Isolation Forest) to flag user behaviors that significantly deviate from the norm and may indicate insider threats.
This project uses unsupervised text modeling techniques to explore online reviews of Nike pr oducts to extract actionable insights for marketing and product development.
The goal of this project was to build an image classification model to recognize facial expressions using deep learning. The dataset contained face images categorized by emotion labels, and the project aimed to classify each image into its respective emotional class. A high-performing ConvNeXt architecture was used to achieve this.
A fine-tuned a transformer based text classifier using the ktrain wrapper over TensorFlow/Keras and HuggingFace transformers was trained to classify text based news articles for targeted marketing in the health & wellness niche.
Binary classification on histopathologic scans; pipeline, metrics, and a clean Kaggle-ready submission.
Supervised learning over MITRE-labeled network data to highlight command-and-control patterns.
Stepwise regression linking environmental and genetic factors (5-HTT) with depression outcomes.
Feature engineering from pitch-level data to predict run-risk innings with ~83% accuracy.
Training a CycleGAN to translate real-world images to Monet-style outputs; data, training, and MiFID.
Transformer-based classification for disaster vs. non-disaster tweets with minimal preprocessing.
Dimensionality reduction + supervised/unsupervised models to categorize news at scale.
From feature selection to evaluation — a practical model to filter low-quality job listings.
Finding revenue-maximizing price points by product with historical demand and price variation.
Transformations, lack-of-fit testing, and why the power model won — a compact walkthrough.