Artificial Intelligence & Machine Learning

Case Studies

A principal-level ML system design case study for proactive, pre-publish harmful content detection on a social platform. Covers requirements and latency SLOs, a rules-to-ML inference cascade, threshold-driven enforcement via a decoupled policy engine, model/version auditability, monitoring and drift detection, and feedback loops for continuous improvement—with extensibility notes for future image/video moderation.

Projects

Here are my applied AI and machine learning projects, focused on real datasets, model comparison, experimentation, and performance tradeoffs. These use tools like pytorch, tensorflow, scikit, and are written in jupyter notebook format.

A full end-to-end notebook demonstrating data preprocessing, model training, optimization tuning, and performance comparison between an MLP and CNN using the MNIST dataset.

An end-to-end NLP notebook that loads the IMDB dataset with Hugging Face, builds a Bag-of-Words MLP baseline, and then trains a custom Mini-Transformer for text classification. Includes preprocessing, tokenization, training loops, optimization experiments, LR range testing, and a full comparison between classical and modern architectures for sentiment prediction.

An end-to-end tabular machine learning notebook using scikit-learn to classify malignant vs benign tumors from diagnostic features. Covers data inspection, stratified splitting, baseline modeling, model comparison, decision-threshold tuning based on domain tradeoffs, probability calibration, and a final held-out test evaluation.

An end-to-end deep learning notebook using PyTorch to classify urban environmental sounds from raw audio. Covers audio decoding and resampling, log-mel spectrogram feature extraction, CNN baselines, controlled experiments (baseline vs SpecAugment-lite), and detailed evaluation with confusion matrices, class-pair analysis, confidence inspection, and spectrogram-based error analysis.

An Introduction

using python & common python libraries to explore & analyze data: statistics, probabiliy, Percentiles, Moments, Covariance, Correlation, Conditional Probability, & Bayes' Theorem. Also, an introduction to tensorflow.

Linear Regressions, Decision Trees, K-Means clustering, Ensemble learning (bagging & boosting), & building models.