// frequency-domain analysis
I study how the same Fourier transform reveals hidden structure in sound, images, audio, and ground motion — from a guitar body to a scanned document to a seismic sensor.
About
I'm a rising junior at Redmond High School focused on physics, mathematics, and signal processing. My work traces one idea — that decomposing a signal into its frequency components reveals what a raw time-series can't — across acoustics, computer vision, audio forensics, and seismology.
This started with a question about my own electric guitar build: why does it resonate the way it does? That question turned into a research throughline connecting a Celusion Technologies internship, independent research, and a Research mentorship.
Research
Each project applies the same core method — FFT / spectral analysis — to a different physical signal.
Detecting spatial-frequency artifacts left behind by copy-paste and re-compression tampering in scanned ID documents, built during a fraud-detection internship at Celusion Technologies.
Investigating which interpretable frequency-domain features best distinguish synthetic from genuine speech, using the ASVspoof dataset — an alternative to black-box detection models.
Measuring the resonant frequencies of a hand-built electric guitar using phone accelerometer and microphone data, comparing measured harmonics against theoretical string physics.
A Research project testing whether frequency-domain signatures in low-cost accelerometer data can flag slope-instability precursors — motivated by the 2014 Oso landslide in Washington State.
The throughline
Each project reapplies the same frequency-domain method to a signal with higher real-world consequence than the last.
Where the question started — personal curiosity.
Applied to fraud & security at Celusion.
Applied to AI-driven fraud, independently.
Applied to public safety, with research mentor.