Hardware + human study
Fine-tuning wearable sensor hardware and conducting experimental human studies.
- Mechanical design, CAD, and 3D printing
- Sensor placement and mounting iterations
- Cable routing and experiment setup
- Human-study acquisition
Cornell University ECE
Hi, myname isYoungseo.
I’m an Electrical and Computer Engineering student at Cornell University exploring wearable biosensors, bioelectronics, signal processing, and machine learning.
LinkedInMicrosoft Student Ambassador
Featured research · Cornell ECE
Wearable Radio-Frequency Muscle Sensing
Cornell University · Edwin C. Kan Lab
Radiomyography · Near-Field Coherent Sensing · Silent Speech Recognition
Can we recognize speech without listening to sound?
At Cornell’s Edwin C. Kan Lab, I worked on a wearable radio-frequency sensing system that captures subtle facial and neck muscle motion using Radiomyography (RMG) and Near-Field Coherent Sensing (NCS). My work spanned wearable hardware optimization, human-subject experiments, RF signal acquisition, signal processing, and machine-learning-based word classification.
Hardware and data,
developed together.
Signal processing + ML
Processing multi-channel RF signals and developing the classification pipeline.
- Synchronization, filtering, and segmentation
- STFT and time/frequency-domain analysis
- Feature extraction and multi-channel representation
- Machine-learning classification
Hardware Signal Features Classification
Speech has a physical signature.
Microphone
Acoustic pressure
- Ambient noise
- Privacy concerns
- Requires audible speech
sEMG
Surface electrical activity
- Skin contact and impedance
- Placement sensitivity
- Electrode alignment
IMU
Motion + orientation
- Gravity coupling
- Orientation drift
- Head-motion artifacts
RMG / NCS
Near-field RF sensing
- Muscle/tissue dielectric change
- Facial + neck sensing
- Noise-resistant interface potential
Four regions.
One coherent RF system.
- RF TxTransmit field
- Near-field electromagnetic coupling
- Facial / cervical muscle tissueMuscle deformation + dielectric modulation
- RF RxReceive the modulated signal
- Measured RMG signalSpeech-relevant muscle activity over time
Build. Fit. Measure. Revise.
Each physical prototype informed the next mechanical and electrical decision.
- CAD
- 3D print
- Prototype
- Fit test
- Sensor integration
- Hardware revision
- Final wearable
CAD + slicing
Translate mounting geometry into a printable part and inspect the Ultimaker toolpath.
3D print + prototype
Check the printed channels, wall geometry, and room for the sensing hardware.
Fit + sensor integration
Integrate the sensor and test fastening and cable clearance on the physical mount.
Revision + final wearable
Bring the revised mounts, headband, cables, and RF connections together at the bench.
From the wearable
to synchronized RF data.
- Facial / neck sensor array
- Tx / Rx RF channels
- SMA connections
- Synchronized USRP radios
- LabVIEW
- Raw multi-channel RMG data
- Gain calibration
- Channel synchronization
- Real-time signal inspection
- Audio reference
Different words leave
different muscle signatures.
Low-frequency NCS, high-frequency NCS, and the audio reference reveal different aspects of each articulation.
The work, documented.
Original poster and available technical-report pages.
Industry · Engineering · Research
Internship Experience
Selected industry work across bioelectronics, pharmaceutical analytics, semiconductor verification, and machine learning.
01 / Semiconductors
LX Semicon
Machine Learning & Circuit Verification Engineer Intern
Team / TV T-Con DDR Part
- Focus
- Display-driver IC reliability
- Work
- Circuit-verification protocols
- Machine-learning evaluation
- Test automation
- Hardware reliability
Designed circuit-verification test protocols and developed machine-learning-assisted evaluation pipelines to reduce repetitive manual checking in display-driver IC reliability workflows.
My work connected circuit validation with data-driven evaluation, making reliability testing more systematic, repeatable, and efficient.
- Circuit verification
- Machine learning
- Display driver IC
- Reliability testing
02 / Industry
ARIBIO
Computer Engineer / Pharmaceutical Analyst Intern
- Focus
- Alzheimer’s drug development
- Work
- Computational modeling
- Statistical data analysis
- Protocol standardization
- Neural-interface concepts
Applied computational modeling and data analysis to Alzheimer’s drug-development research, supporting preclinical analysis through statistical data collection and protocol standardization.
Also contributed to neural-interface concepts for Neuro:Re, exploring data-driven approaches to neurodegenerative disease diagnostics.
- Computational modeling
- Pharmaceutical analytics
- Alzheimer’s research
- Neurotech
Research & Engineering
01 / Project team
Engineering World Health
Integrative Design Engineer · Cornell
Team websiteIn collaboration with Solar Ear, we are developing an innovative, kinetic-energy, body-worn “Action Figure” hearing aid aimed at expanding access for children with hearing loss in low-resource settings. I contributed to the technical development by designing the AutoCAD double pendulum model for the kinetic energy system and designing the relevant circuit board. I also had the opportunity to present the kinetic energy pendulum model at the Boston Consortium for Global Health and Innovation (CGHH) Conference and led a specialized electrical PCB analysis workshop for team members.
Our current work centers on vital global health projects in collaboration with “2ft Prosthetics” and “Solar Ear.” For 2ft Prosthetics, I am contributing to the development of the OneSize Foot, a high-priority, low-cost prosthetic engineered for adaptability in resource-limited settings.
AutoCAD double pendulum · PCB design · Global health02 / Research
UCLA2026 - Present
Bioelectronics Researcher
Jun Chen Lab Team website
Researching rheologically adaptive magnetic bioelectronics for motion-robust throat-based speech decoding. An ultrastretchable interface is being evaluated with NATO phonetic-word recognition.
Can the interface preserve speech-relevant laryngeal biomechanics while reducing motion interference before the signal reaches the electronics?
CornellMay 2025 - Present
Hardware & Software Engineer
Edwin C. Kan Lab Team website
Designed and fine-tuned wearable RMG sensing hardware using near-field coherent sensing, and developed signal-processing and machine-learning workflows for facial and neck muscle classification.
88.09% classification accuracy · View research03 / Additional engineering
Software Developer
Engineers for a Sustainable World · Cornell
Redesigned the organization’s web experience, improving layout, interaction, accessibility, and digital infrastructure.
HTML / CSS / JavaScriptBioelectrical Researcher
Jellyfish Sweat Biosensor · Independent research
Developed a semipermeable jellyfish-skin-inspired sweat patch, improving electrolyte permeability through iterative experimental testing.
300+ trials · +80% permeabilityMicrosoft Student Ambassador
Microsoft
I’m a member of the Microsoft Student Ambassadors community.
Community, writing & outreach
TEDx speaker mentorship and event leadership; founding The Women Engineers International and the Thornton Academy STEM Club; STEM tutoring; and writing Mathematical Answers to Human Existence.
Other projects
Questions I’m building around.
Adaptive Magnetic Bioelectronics
Can a material filter motion before the electronics do?
An ultrastretchable interface for throat-based speech decoding.
Explore
At Jun Chen Lab, I’m investigating whether a rate-dependent interface can preserve speech-relevant laryngeal biomechanics while reducing motion-induced interference before acquisition and machine-learning analysis.
Jellyfish Sweat Biosensor
How can a biological membrane improve wearable sensing?
A semipermeable sweat patch developed through more than 300 experimental trials.
Explore
Iterative testing improved electrolyte permeability by 80%. I presented this work at the 2023 NANO Korea Symposium, exploring how the membrane could support wearable sweat diagnostics.
Solar Ear + OneSize Foot
How do assistive devices adapt to everyday constraints?
Hearing-aid concepts for off-grid access and mechanical refinement of a split-toe prosthetic.
Explore
With Solar Ear, I worked on kinetic-energy-powered hearing-aid concepts centered on children. For OneSize Foot, I optimized the split-toe design for durability and stability, supporting customized fittings for 18+ individuals, including recipients in Tonga.
Engineering for Community
How can a student organization’s website work better?
A redesigned web experience for Cornell’s sustainable engineering community.
Explore
I restructured the organization’s website architecture, refined page layouts, and developed interactions with attention to accessibility and performance.
Skills
From physical systems to useful signals.
Hardware + electronics
- Circuit design
- CAD design
- Wearable sensors
- RF / wearable sensing
- Prototyping
Signals + ML
- Python
- Machine learning
- MATLAB
- Signal processing
- Wearable signal processing
Software
- C / C++
- HTML / CSS
- JavaScript
- Next.js / React
- Web development
Data + engineering
- Excel
- Data analysis
- Experimental design
- Protocol development
Let’s build something interesting.
I’m always interested in wearable sensing, bioelectronics, hardware–software systems, and ambitious engineering problems.
About
Engineering, but softer around the edges.
I like building systems that make invisible signals tangible, from wearable sensors and 3D-printed hardware to circuits, signal-processing pipelines, and machine-learning models. I’m especially interested in the space where electronics meet the human body, and where careful engineering can turn noisy biological information into something useful.
My work moves between hardware and software. Sometimes that means redesigning a sensor mount after a noisy experiment; sometimes it means tracing a circuit-verification workflow or building a classifier from physiological signals. I like being close enough to the hardware to understand where the data came from.