AI & Machine Learning
Predictive modeling, deep learning, NLP, and information retrieval, backed by applied AI projects.
My AI and machine learning work combines data preparation, model development, and evaluation with a software engineering foundation. These examples are independent personal projects, rather than AI work attributed to an employer.
Technical Focus
Python, PyTorch, Scikit-learn, Pandas, NumPy, MATLAB, C++17
Personal Project — NASA C-MAPSS Predictive Maintenance
AI (Machine Learning) Engineer · January 2025–Present · Personal project portfolio
Work Delivered
Developed aircraft-engine remaining useful life prediction workflows using NASA C-MAPSS sensor data. Built reusable preprocessing for normalization, feature selection, noise reduction, and sliding-window sequences, then implemented MLP, LSTM, and temporal convolutional network models.
Achievements & Outcomes
- Created configurable training and evaluation workflows for FD001–FD004, with model checkpoints and performance visualizations.
- Designed an asymmetric loss function that penalizes remaining-life overestimation to reflect predictive-maintenance risk.
Personal Projects — Classification & Natural Language Processing
Machine Learning & NLP Development · 2025–Present · Personal project portfolio
Work Delivered
Applied label distribution learning to tennis performance classification; extended an email classification application to distinguish spam, ham, phishing, and harmful content; and developed an automated essay-grading system using writing features, Lasso Regression, and Decision Tree models.
Achievements & Outcomes
- Compared five statistical learning methods for tennis performance using classification error, KL divergence, and cosine similarity.
- Combined probability outputs from two email classifiers to produce ensemble labels and confidence scores.
- Evaluated essay-score prediction with MAE, MSE, and R² across K–12, college, and graduate-level writing categories.
Personal Projects — Information Retrieval & Image Watermarking
C++17, Python & MATLAB Development · January 2025–Present · Personal project portfolio
Work Delivered
Built a C++17 information retrieval engine with parsing, tokenization, stopword removal, Porter stemming, and forward and inverted indexes. Developed a separate image-watermarking application using Haar discrete wavelet transforms in MATLAB and Python.
Achievements & Outcomes
- Implemented document-term indexing and term-frequency tracking for efficient term lookup.
- Supported watermark embedding, extraction, and verification, with PSNR, SSIM, BER, and NCC evaluation.
How I Approach This Work
Start with the prediction or retrieval problem, prepare reproducible data pipelines, compare suitable baselines, and evaluate with task-specific metrics. Keep training configurations, checkpoints, and visualizations reusable so experiments can be inspected and repeated.
