— Projects
Commonwealth HR Assistant
Led human-centered design discovery with HR stakeholders to scope a RAG-based AI assistant projected to benefit 6,300 employees, then built and evaluated a hybrid BM25 + dense retrieval pipeline on AWS Bedrock Knowledge Bases — achieving roughly 15-second response times and a 90%+ user satisfaction rate. Diagnosed and resolved retrieval-quality issues across 5 collective bargaining agreements, including table-chunking failures and cross-document retrieval bleed.
next.js · fastapi · aws bedrock · hybrid retrieval
MRI Tumor Detection
Engineered a CNN with dual global pooling (average + max, concatenated) to classify brain MRI scans into four tumor types, using Grad-CAM to diagnose a background-shortcut-learning bias and eliminate it via content-aware cropping and random-erasing augmentation. Achieved 92% accuracy on a 1,600-image held-out test set; deployed via FastAPI and Docker on Google Cloud Run with CI/CD and Hugging Face Hub-based model versioning.
tensorflow · keras · cnn · opencv
> more in training...
— About
I grew up in South Africa, where computers fascinated me from a young age. Throughout my childhood, I loved learning everything I could about them. I've always enjoyed finding out new things about them and trying to understand it, and that curiosity only grew stronger over time.
To pursue that passion, I was fortunate enough to move to the United States to study it, and after a lot of hard work, I made my way to UMass Amherst, a school renowned for its work in AI and machine learning.
Now studying Data Science and ML Engineering there, I'm building the technical depth to match the passion I've had since I was young, and I'm loving every second of it. My goal is to grow from here into a career as an AI/ML Engineer, doing real work that makes a difference, and I'm hoping to continue on to a Master's in Computer Science along the way.
- Python
- TensorFlow / Keras
- XGBoost
- scikit-learn
- pandas / NumPy
- SQL
- Docker
- AWS
- Google Cloud Platform
- CI/CD
- Git
- Statistics & Probability
- Agentic Coding (Claude Code, Kiro)
— Resume
A snapshot of my background, skills, and trajectory.
— Academics
0.0
/ 4.0
Cumulative GPA · UMass Amherst
Relevant Coursework
| Course | Grade |
|---|---|
| Machine Learning | A |
| Statistics | A |
| Algorithms | A |
| Simulation Modeling | A |