Hello, I'm Andrew Ma

Computer Science Student at UT Austin (Class of 2029) | AI & Machine Learning Researcher.

Bridging the gap between theoretical research and practical application. From co-authoring papers on LLM efficacy to building deep learning engines from scratch, I am driven to understand the "black box" of AI.

View Work Contact Me

Technical Arsenal

Languages

Python (Advanced) Java C++ JavaScript (ES6+) SQL

AI & ML

Deep Learning TensorFlow PyTorch NLP (BERT, Llama) Computer Vision LoRA

Web & Cloud

React.js Node.js AWS (EC2, S3) RESTful APIs CI/CD

Data Science

Pandas NumPy SHAP Visualization

Experience

Oct 2025 – Present

Research Assistant: DeepFake Detection

UT Austin Dept of Computer Science | Lab Director: Dr. David Harwath

Engineering robust detection mechanisms for audio and text-based DeepFakes.

  • Architected a high-accuracy NLP classification pipeline (TF-IDF + BERT).
  • Utilized Ensemble Learning to aggregate predictions, achieving >85% accuracy.
  • Conducted linguistic artifact analysis to identify subtle model-generated patterns.
May 2024 – Aug 2024

Co-Author: Opinion Mining with AI

Lamar University | Advisor: Dr. Rakibul Islam

Benchmarked modern generative AI against traditional transformer models for Opinion Mining.

  • Stressed-tested architectures including Llama, GPT-4, and RoBERTa.
  • Led implementation of LoRA for efficient fine-tuning.
  • Co-authored a paper on lightweight transformers vs. massive generative models (Read Paper).
Apr 2023 – July 2023

Development Assistant: Renewable Energy Robotics

Lamar University | Advisor: Dr. Stephan Andrei

Modernizing solar-powered robotic units for energy efficiency.

  • Rewrote control logic for RTOS integration.
  • Developed algorithms for dynamic power consumption adjustment.
  • Improved physical build with higher-yield solar cells.

Featured Projects

Stroke Prediction: Advanced ML Pipeline

Healthcare AI | Imbalanced Data | Explainability

Addressed early stroke detection using patient health data with a Stacking Ensemble model (Random Forests + Neural Networks).

  • Implemented SMOTE for minority class over-sampling.
  • Integrated SHAP values for model explainability.

"From Scratch" Series: The Math Behind the Code

Algorithms | Calculus | Core Python

Implementing core ML algorithms entirely from scratch to understand the engine under the hood.

  • Micrograd Extension: Built a modular Neural Network engine with Autograd.
  • Linear Regression: Wrote raw cost function analysis and Gradient Descent optimization loop.

Leadership & Impact

Beaumont CS4ALL

Founder & President

Founded a non-profit to democratize CS access. Scaled to 30+ participants and designed a Java curriculum for local libraries.

Byte Club

Founder & President

Established the first CS club at West Brook High School for collaboration and competitions.

Philanthropy

Tech Drive & Tutoring

Raised $3,000+ in devices. Coached FPS teams to state finishes. Tutored math/reading at local elementaries.

Awards & Recognition

  • 🏆 USACO: 1st Place & Perfect Score (Bronze & Silver)
  • 🏅 UIL Computer Science: 4th Place (District 2024)
  • 🔬 TMSCA State Science: 1st Place (2022)
  • 🧠 FPS: 1st Place State Champion (2023)
  • 🎓 Scholastic: AP Scholar with Distinction, NMSI Scholar