Thomas Boziuk Email & Phone Number

Predictive Analytics: Traumatic Brain Injury Recovery • MVP Prototype Design: Initial ensemble model development on historical data. • Identified key parameters via Shapley value explanations and pruned required features to collect by 80% while maintaining average performance goal of 0.8+ AUC. • Serverless backend design on GCP (Firebase, BigQuery, Cloud Functions/Run) for iOS app. • Orchestrated data collection (user input and wearable devices), processing, and display. • Causal inference modeling for treatment effects of potential interventions.Computer Vision: Object Tracking for Medical Inventory Management • Managed dataset ingestion, annotation, and outlier detection to identify future image acquisition requirements. • Doubled detectable object classes while increasing target metric of F1 score by 10% via hyperparameter tuning and adding image augmentation pipeline.

Atlanta, Ga

• Investigated Keras-based convolutional neural network image segmentation of vapor bubbles. • Designed application-driven experiments involving multi-scale acoustic actuation for the enhancement of heat transfer in industrial process engineering. Collaborated with other universities for CFD modeling and an industry trade group to scale technologies for pilot-scale field test. • Implemented encoder-feedback control and sensor safety system of hermetically-sealed linear actuator for perturbing pressurized R-134a flows at controlled phase mixtures for heat transfer enhancement measurements via sensitive energy balance method. Demonstrated 30% condensation enhancement at low quality, stratified-flow mixtures. • Designed experiment to investigate laser-induced cavitation bubble collapse and the ability of acoustic actuation to affect the directionality of the collapse process near solid boundaries to reduce cavitation damage.

Thermodynamics (ME3322) and Heat Transfer (ME3345)

• Performed data acquisition, processing, statistical analysis, error estimation and visualization. • Developed Python-based image processing to estimate heat transfer enhancement of vapor bubble condensation. • Designed experiments to investigate interactions between multi-scale acoustic actuation and vapor-liquid heat transfer interfaces. Investigated applications based on the fundamental phenomena. • Enhanced condensation of vapor bubbles in subcooled fluid by capillary surface waves by 255 percent. Designed and fabricated custom housings for piezoelectric actuator to allow submerged operation at ambient and reduced system pressures. Determined fundamental cause of condensation enhancement via flow visualization experiments. • Enhanced condensation of vapor bubbles by 355% using an ultrasonic beam that deforms the bubble, induces droplet ejection, and leads to a rapid collapse. Demonstrated feasibility under varied geometric conditions, and developed efficiency gains by adding a synchronization system to detect bubble formation and subsequently pulse actuation. • Increased critical heat flux in pool boiling by 65% using ultrasound actuation of vapor-liquid interfaces. Boiling effectiveness was further increased by coupling the acoustic enhancement with custom-fabricated micro-textured surfaces.

Doctor Of Philosophy (Ph.D.), Mechanical Engineering

Acoustic Actuation of Vapor-Liquid Interfaces in Boiling and Condensation Processes

Master Of Science (Ms), Mechanical Engineering

Bachelor Of Science (Bs), Mechanical Engineering