SebastianBuxman
Johannesburg to California
Robots, remittances, tennis.
About
I study how people and collaborative robots can work side by side without the person paying for it physically.
I'm a second-year Mechanical Engineering PhD student at Santa Clara University, in the Machine Learning and Safety Analytics Lab. Before that I did a bachelor's in Computer Science at Weber State University and a master's in Computer Science and Engineering at Santa Clara.
I came to engineering through software and to the United States through tennis. I'm also co-founder of Kwama, a remittance startup for the Zimbabwe–South Africa corridor. Outside of work: tennis, amapiano, and long runs.
Research
My work uses wearable sensor data and machine learning to measure what a task is doing to a person's body, then feeds that back to the robot. The methods are deep learning on muscle and motion signals and fault tree analysis; the goal is a closed loop where the robot adapts to the person in real time.
Papers
Four peer-reviewed papers, all first author.
-
Deep Learning–Based Temporal EMG Modeling for Task-Specific Deployment of Passive Exoskeletons in Manufacturing Settings
S. Buxman, F. Davoudi Kakhki. Models muscle activity over time to decide which manufacturing tasks a passive exoskeleton should be used for.Flexible Automation and Intelligent Manufacturing, Paris / Springer LNME
FAIM 2026 / In press -
Multimodal Evaluation of Exoskeleton Assistance Using Optimized Machine Learning
S. Buxman, et al. Combines muscle activity, motion and self-reported workload to classify lifting with and without an exoskeleton.Human Factors and Ergonomics Society Annual Meeting
HFES 2026 / Published → -
Breakpoint Modeling of Surface EMG-Derived Fatigue Onset for Human-Robot Collaborative Assembly Under RULA-Differentiated Postures
S. Buxman, et al. Piecewise regression on muscle signals to find when fatigue starts. Ergonomic posture delays biceps fatigue onset about threefold.ASME Manufacturing Science and Engineering Conference
MSEC 2026 / Published -
Sensor-Driven Strain Detection and Deep Learning Evaluation of Passive Exoskeletons in Industrial Tasks
S. Buxman, et al. Wearable sensor data from back-support and shoulder exoskeletons. Exoskeletons redistribute load across muscles; they do not reduce it uniformly.IEEE/SICE International Symposium on System Integration
SII 2026 / Published →
Ventures
Projects
-
Ball Balancing Platform
An overhead depth camera and two servo motors keep a ping-pong ball centred on a tilting plate, or steer it around circles, squares and figure-eights. Built with Shaunak Sharma for Mechatronics II.Python / OpenCV / PID control / Arduino / OAK-D Pro / MATLAB
Code → -
Load Shedding Power Hub
A peer-to-peer solar energy marketplace for South Africa.React / Node.js / AWS / DynamoDB
Code → -
ML on Tennis Players
ATP player data analysed to find the attributes that predict a top-100 ranking.Python / Pandas / Scikit-Learn / Matplotlib
Code → -
IoT Smart Fridge Tracker
Raspberry Pi, sensors and image classification that track what is in the fridge and when it expires.Raspberry Pi / OpenCV / Flask / AWS / Twilio
Code → -
Serverless Dynamic Pricing
A pricing engine that adjusts prices to real-time demand.AWS Lambda / DynamoDB / Python
Code → -
Credit Card Fraud Detection
Models that detect fraudulent transactions in highly imbalanced data.Python / Scikit-Learn / XGBoost / SMOTE
Code → -
Design Patterns in a Costco System
18 design patterns implemented across bakery, stock, customer support and other subsystems of a simulated Costco.Java / OOP / Software architecture
Code →
Tennis
I captained the team at Weber State, then played for Santa Clara during my master's. I was also a graduate assistant coach for Santa Clara's Division I team.