Resume

University of Texas at Austin

Bachelor of Science in Biomedical Engineering, Computational Track

Relevant Coursework: Embedded Systems, Biomedical Instrumentation, Circuits, Systems & Signals, Differential Equations & Linear Algebra, Intro to Computational Engineering Design, Statistics, Numerical Methods

Edge AI Team Lead

Longhorn Neurotech · Austin, TX

• Directed a team of student engineers to optimize neural networks for deployment on embedded systems (Raspberry Pi series).
• Designing a reusable quantization pipeline using ExecuTorch and an inference backend to support keyword spotting and future neuroprosthetic control applications.
• Developed an optimized ONNX inference pipeline to process EMG band signals for prosthetic arm control, reducing MLP model energy consumption by 45%.

AI/ML Developer

Longhorn Neurotech · Austin, TX

• Developed machine learning models for Brain-Computer Interfaces (BCI) aimed at controlling a prosthetic arm, using architectures such as CNNs and Capsule Networks.
• Improved BCI model accuracy from 60% to 80% by tuning hyperparameters with Optuna and using Captum to reduce overfitting.

Machine Learning Researcher

Wang Laboratory, UT Austin · Austin, TX

• Gut-Brain Axis Research: Writing machine learning models to analyze EEG and EGG data, predicting the optimal timing for targeted ultrasound vagus nerve stimulation to help alleviate gastrointestinal issues.
• Neurawear Startup Initiative: Collaborating with a student engineering team to develop the "Neustim" wearable device, processing epilepsy datasets to build predictive neuromodulation algorithms.

Researcher and Programmer

Functional Optical Imaging Lab, UT Austin · Austin, TX

• Developed a real-time Laser Speckle UI in Python, using Basler cameras and Arduinos for automated laser intensity control.
• Enhanced speckle pattern evaluation by implementing contrast analysis algorithms to maximize measurement precision.
• Refactored contrast adjustment scripts from MATLAB to Python, enabling integration with the real-time Laser Speckle Analysis UI.

MedAdvisor — Clinical Feedback App

• Built an iOS app with a Stanford surgeon that records medical consultations and scores them against a clinical rubric on-device.
• Deployed a 4-bit 7B LLM on iPhone (llama.cpp, Metal GPU): full rubric analysis in under 4 minutes, running the 4 GB model in a 495 MB footprint via memory-mapped weights.
• Cut rubric evaluation latency ~60% (18.6 s → 7.3 s/criterion) by caching transcript KV state across all 16 evaluations; built an on-device benchmark that quantified 31% thermal throttling under sustained load.

Cloud to Edge Heart Rate Monitor

• Architected a cloud-to-Raspberry Pi 5 data pipeline streaming real-time heart rate data for patient simulation.
• Developed a UART communication protocol to stream simulated patient data from the Pi 5 to a Pi Pico for low-latency processing.
• Built a custom C-based inference engine on the Pico's dual-core Cortex-M0+ architecture to execute quantized neural networks on the incoming data stream in order to classify the patient's heart condition.

Embedded Multiplayer Racer (ARM Cortex-M0+)

• Earned 'Best Embedded Design' in a class-wide competition for excellence in hardware-software integration.
• Built a bare-metal multiplayer racing game in C for a TI MSPM0 microcontroller, implementing collision detection, velocity-based physics, and sprite rendering.
• Programmed register-level drivers for UART, ADC, and DAC to interface with analog joysticks and audio peripherals, bypassing standard libraries for efficiency.
• Designed a custom UART protocol to synchronize player coordinates and game events between two microcontrollers in real-time.