At the Harvard Biodesign Lab, designed, fabricated, and developed wearable ankle exoskeleton robots to assist gait. Created detailed designs in SOLIDWORKS and fabricated multiple robot prototypes using SLS 3D printing, carbon fiber molding, and soldering new electronic components. Conducted comprehensive data collection and analysis using MATLAB and Simulink, incorporating both on-body and bench-top testing to optimize performance through power flow analysis and iterative design methods.
Me walking with device:
This work contributed to a paper featured at the IEEE International Conference on Robotics and Automation.
Captured a high-fidelity dataset to design and validate custom 3D human pose tracking algorithms, compare IR, RGB, and point-cloud data quality between different camera alternatives, and perform validation assessment of new solutions with respect to a reference motion capture system. Designed and implemented a novel automated anomaly detection method for motion capture data to improve tracking reliability. Performed statistical analysis on the resulting datasets to validate model performance.
Established a transcranial magnetic stimulation (TMS) experimental setup for investigating paired associative stimulation (PAS) protocols in stroke rehabilitation research. Independently integrated hardware and software systems using Cambridge Electronic Design (CED) Signal software and MATLAB to control stimulation parameters and data acquisition. Conducted extensive data analysis on noisy experimental datasets to extract meaningful insights for clinical applications. Below are two posters detailing the setup and the research details.
As a course project, co-designed and optimized a lightweight aerospace gearbox based on Solar Impulse aircraft specifications as part of a four-member engineering team. The gearbox was required to withstand 60 HP at 5,500 RPM while transmitting power to a propeller at 835 RPM, meeting strict aerospace requirements for weight, reliability, and safety factors. Applied AGMA and ASME standards to ensure all components exceeded the required 1.5 safety factor through iterative optimization, stress analysis, fatigue life estimation, advanced material selection, and CAD modeling. Delivered a final design with manufacturing-ready CAD drawings for the complete gearbox assembly.
Developed an algorithm to verify as-built footing geometry using LiDAR point cloud data, integrating Building Information Modeling (BIM) principles to ensure construction accuracy.
Leveraged advanced computational methods—including Singular Value Decomposition (SVD), modified K-means clustering, and plane intersection—to process complex datasets and extract key geometric features such as planes and corner points.
Built comprehensive 3D analysis workflows to compare as-built structures with design specifications, demonstrating proficiency in large-scale spatial data processing, computational geometry, and construction technology.
