Yuta Kiami

Engineering • Physics • Computation

About Me

I'm a third-year student at UCLA studying Physics and Bioengineering, working at the intersection of computational methods, engineering design, and the natural sciences. I enjoy learning and thinking at the scientific level while applying my knowledge and skills to solve real-world problems, whether that's better understanding the physics of antimicrobial peptides with ML, or restoring music playing for those who have lost hand mobility. Looking forward, I hope to grow my technical experience through industry roles with teams building something new. Outside the lab, I play cello, enjoy cooking, and spend as much time as I can skiing or outdoors.

Projects

SSVEP Brain-Computer Interface Piano Player

BCI EEG FBTRCA Python Signal Processing UI/UX Real-Time Classification
  • Built a brain-computer interface system allowing users to play piano notes hands-free by gazing at flickering visual stimuli (SSVEP).
  • Developed an integrated UI combining an EEG data collection interface and a live classification display, enabling seamless transitions between training and performance modes.
  • Implemented and trained Filter Bank Task-Related Component Analysis (FBTRCA), a state-of-the-art SSVEP decoding algorithm, achieving robust multi-class frequency discrimination.
  • Mapped decoded SSVEP frequencies to piano notes, triggering audio playback with EMG to create a fully functional BCI musical instrument.
  • Iterated filter bank parameters and epoch lengths to maximize accuracy.
CruX Competition Presentation →

Epistasis in Plasmid Horizontal Gene Transfer

Evolutionary Biology NK Landscapes Gillespie Algorithm Monte Carlo Python Stochastic Simulation Epistasis Numba NumPy Pandas
  • Investigating how epistasis, non-additive interactions between plasmids, shapes the dynamics of horizontal gene transfer (HGT) via plasmids in bacterial populations.
  • Modeling fitness landscapes using the NK framework, tuning the epistasis parameter K to produce landscapes ranging from smooth and additive to highly rugged and multipeaked.
  • Simulating population-level evolutionary dynamics using the Gillespie and Monte Carlo algorithms for stochastic kinetics, capturing birth, death, mutation, and plasmid transfer events at the individual level.
  • Systematically iterating over landscape ruggedness, transfer rate, and population size parameters to map out how HGT interacts with epistatic structure to facilitate or impede adaptation.
  • Probing whether plasmid-mediated gene transfer acts as an escape mechanism from local fitness peaks on rugged landscapes, and under what epistatic regimes HGT is most evolutionarily advantageous.
Epistasis in Plasmid HGT simulation

TruBalance Stand

Constraint-Driven Design Jupyter SolidWorks Prototyping 3D Printing
  • Creating a rigid tablet stand defined by a curve that rests at static equilibrium for any angle at which the stand can be placed.
  • Considered physical and geometrical constraints to work out an initial design concept, reducing to only two constraint equations.
  • Used Jupyter to numerically solve for constrained parameters, allowing for easy material and dimensional customizability.
  • Modeled the design in SolidWorks to allow prototyping.
  • Currently creating prototypes using 3D printing, making practical and design improvements with each iteration.
Documentation →

Peptide Curvature VAE Pipeline

VAE KNN Julia Bash Dimensionality Reduction Bioinformatics
  • Automated a pipeline to splice peptide sequences into overlapping segments of varying lengths, enabling sub-sequence level analysis of membrane curvature affinity.
  • Passed spliced segments through a Variational Autoencoder (VAE) to produce low-dimensional embeddings encoding the physical and biochemical properties of each peptide fragment.
  • Applied a K-Nearest Neighbors classifier in the VAE latent space to assign each segment a negative Gaussian curvature (NGC) or positive Gaussian curvature (PGC) affinity score based on proximity to labeled reference peptides.
  • Visualized classification results across peptide length to provide interpretable, position-resolved curvature profiles for downstream analysis.
  • Automated the full workflow end-to-end using Julia and Bash scripting, enabling rapid throughput across large peptide libraries.
Peptide curvature hotspot visualization

Flagellum Image Classification

CNN Julia Pluto Image Processing Bash
  • Built a deep learning pipeline to classify 10,000+ bacterial images with high throughput and reproducibility.
  • Designed a streamlined interface for manual labeling, enabling 600+ images/hour for training data generation.
  • Tuned model architectures to boost accuracy and training efficiency, achieving 50%+ accuracy improvements.
  • Experimented with diverse datasets and processing, boosting model accuracy by 30%.
Flagellum mosaic

EEG Seizure Detection

DWT Jupyter SciPy SVM PCA
  • Used machine learning methods to create software that automatically classifies EEG as seizure/non-seizure.
  • Processed the signals two ways: one using DWT (discrete wavelet transform) and the other using PCA.
  • Used SVM to classify EEG signals, achieving 96% accuracy, 90% precision, and 90% recall.
Documentation →

TCR-β Design Pipeline

Jupyter Bash AlphaFold OmegaFold ESMFold Pandas NumPy Matplotlib
  • Partnered with a startup and applied my method to design a linker for a TCR-beta structure that could create drug delivery that is more specific, and navigates the body with greater efficiency.
  • Developed a method to computationally design and evaluate 1300+ protein linkers for TCR complexes using AlphaFold, ESMFold, RoseTTAFold, and OmegaFold, facilitating in vitro testing.
  • (Reach out for GitHub repository.)
PyMOL protein alignment example
Download Full Resume (PDF) →

Education

University of California, Los Angeles June 2027
Physics B.A. & Bioengineering B.S. GPA: 3.94 / 4.00
Computer Science Machine Learning Transducers Biomolecular Material Science Mass Transport Thermodynamics Quantum Mechanics

Technical Skills

Languages
Python C++ MATLAB Julia Bash
Python Libraries
NumPy Pandas SciPy Numba Matplotlib Jupyter PyLSL
Machine Learning
CNN DNN VAE SVM PCA KNN FBTRCA
Signal Processing
FFT DWT Filter Banks Bandpass Filtering Epoch Segmentation
Simulation
Monte Carlo Gillespie Algorithm Stochastic Simulation
Tools
SolidWorks LSL Lab Recorder

Professional Experience

Shenshen Wang Theoretical Research Group Los Angeles, CA
Undergraduate Researcher May 2025 – Present
Python Numba JIT NumPy Pandas Matplotlib Monte Carlo Gillespie Algorithm NK Landscapes Stochastic Simulation
  • Independently designed a model of plasmid HGT on NK landscapes to quantify how epistasis drives persistence.
  • Built a Python simulation pipeline and achieved 50x speedup via Numba JIT, enabling large-scale parameter sweeps.
  • Complemented stochastic simulations with analytical derivations to cross-validate results and identify closed-form regimes.
  • Translated findings into visualizations and presentations, partnering with wet-lab collaborators to test predictions.
Gerard Wong Laboratory Los Angeles, CA
Undergraduate Researcher May 2025 – Present
Python Julia Bash Pluto CNN DNN VAE KNN NumPy SciPy Image Processing
  • Built a deep learning pipeline to classify 10,000+ bacterial data images with high throughput and accuracy.
  • Designed a streamlined interface for manual data labeling, enabling 600+ images/hour for training data generation.
  • Tuned model architectures and experimented with image processing methods, achieving a 43% improvement in accuracy.
  • Designed and automated a data analysis and visualization pipeline for peptide fragments using Notebooks and Bash scripts.
UCLA Optofluidics Systems Laboratory Los Angeles, CA
Undergraduate Researcher Nov 2024 – June 2025
PDMS Fabrication Laser Microfabrication Acoustofluidics Thermal Bonding Degassing Single-Cell Manipulation Chip Prototyping
  • Prototyped graphite-PDMS acoustofluidic chips for single-cell manipulation, allowing for efficient iteration of designs.
  • Applied laser microfabrication and automation to produce 60+ microscopic air cavities per minute in graphite chips.
  • Applied a complex fabrication process, including degassing, thermal bonding, and laser manufacturing.
Fred Hutch Cancer Center Seattle, WA
Computational Research Intern Jun 2024 – Sep 2024
Python Bash AlphaFold RoseTTAFold ESMFold OmegaFold Pandas NumPy Matplotlib PyMOL
  • Developed a computational pipeline to design and evaluate 1300+ protein linkers, integrating AlphaFold and RoseTTAFold.
  • Partnered with a startup to apply my pipeline for a novel TCR-β drug delivery system to increase delivery efficiency by 23%.
  • Organized and presented findings to collaborators, including Nobel laureate Dr. David Baker (Institute for Protein Design).

Leadership & Clubs

CruX UCLA Los Angeles, CA
Educational Vice President Mar 2026 – Present
EEG SSVEP FBTRCA Signal Processing Python BCI
  • Design a signal processing, machine learning, and neuroscience curriculum for new members as an introduction to BCI.
  • Formulate a detailed project plan and provide continuous guidance for next year's neuroengineering competition team.
  • Manage and guide a team of 4 technical coordinators in developing and presenting 20+ workshops and technical events.
Neuroengineering Competition Team Engineer Oct 2025 – Apr 2026
SSVEP EMG FBTRCA EEG Signal Processing Python SciPy NumPy PyLSL LSL Lab Recorder Real-Time Classification BCI
  • Designed an SSVEP+EMG neuro-piano for real-time playing, iterating prototypes to achieve 96% classification accuracy.
  • Designed a full-stack pipeline including live signal processing, ML classification, and an intuitive GUI.
  • Won first place at the California Neurotechnology Conference against other schools, including Berkeley, USC, and UCSD.
Technical Coordinator Mar 2025 – Mar 2026
  • Revamped 20+ technical workshops for new CruX members, increasing new member retention by 60%.
  • Advised existing BCI teams and onboarded new teams across software, signal processing, and project workflow.
Samueli School of Engineering Los Angeles, CA
Engineering Ambassador May 2025 – Present
  • Lead 5+ tours/quarter and present engineering programs to prospective students, providing insights and live Q&A.
  • Support admissions and recruiting events as a representative of the engineering student body.