Curriculum Vitae

Vanderbilt student seeking research engineering internship. Research interests are computer vision applied to fields of medical imaging and defense. Experience in deep learning, machine learning, data collection, domain application, and software development. View CV | View Resume

Education

• GPA 3.99/4.00
• Major: Computer Science
• Minors: Applied Mathematics and Data Science

Professional Experience

• Train state-of-the-art pose estimation model on COCO human pose dataset to achieve 98% pose accuracy and brainstorm experiments for pre-seed physical therapy startup as data science hire #3
• Improve pose estimation model for real-time inference by experimenting and choosing most efficient model architecture
• Tech Stack: Jupyter Notebook, Matplotlib, NumPy, OpenCV, PyTorch, Python3, Ubuntu Linux, W&B

• First author, “SpecFlow: A Specular Reflection Restoration Framework using Flow-Guided Video Completion” (Manuscript in progress) - Contact me for a Demo
• Engineered a state-of-the-art ASPP model with optimal hyperparameters to detect specular reflection in endoscopy videos with 92.8% Dice Score and increased sensitivity by 52.3% as compared to previous state-of-the-art U-net models
• Leveraged state-of-the-art optical flow estimation & vision transformer models to restore specular reflection regions with an increased 16.8% PSNR and 10.1% SSIM metrics as compared to previous state-of-the-art spatial inpainting methods
• Tech Stack: Anaconda, Git, Gradio, Hydra, HuggingFace, Jupyter Notebook, Linux Ubuntu, MATLAB, Numpy, OpenCV, Optuna, PyTorch/Lightning, Python3, W&B, YAML

• AirForce GPU-accelerated State Estimation:
  • Designed and implemented backend engine architecture by translating MATLAB code to C++
  • Engineered MVP from scratch in C++ using single GPU to reduce runtime by 21.8% compared to MATLAB
  • Wrote 20+ GPU-accelerated linear algebra utility functions in C++ with corresponding unit tests
  • Tech Stack: C++, CUDA wrapper (MAGMA), GTest, Gitlab, Linux Ubuntu, MATLAB, Matplotlib, Python
• Navy Fooling Classifiers using Adversarial Patterns:
  • Validated 20+ adversarial ML patterns' effectiveness on 30+ state-of-the-art PyTorch classification model
  • Collected 20+ videos of scale vehicle model data from a lightbox camera setup
  • Tech Stack: Anaconda, Jupyter Notebook, Linux Ubuntu, NumPy, OpenCV, Pandas, PyTorch, Python3

• Coordinated metadata collection on a Software-Defined Network using information from 20+ literature reviews to use as a benchmark dataset
• Trained a Proximal Policy Optimization RL model on a custom-made environment to predict the best Azure Files storage tier to place data based on daily size and read/write frequencies
• Optimized file archival process and minimized data storage costs by 74% during backtesting
• Tech Stack: Git, Gym, Jupyter Notebook, Matplotlib, NumPy, Python, SQL, Stable-Baselines3

• Built an explainable AI pipeline to assess prostate tumor detection from mpMRI scans with Logistic Regression, Support Vector Machine, Random Forest, XGBoost, and PyTorch CNN with saliency mapping for each of three prostate zones.
• Published paper “Prostate Lesion Detection and Salient Feature Assessment Using Zone-Based Classifiers” in the Summer STEM Institute Journal, selected as 1 or 5 papers out of 80+ submissions
• Tech Stack: Git, HDF5, Jupyter Notebook, PyTorch, Python, Sci-kit Learn

Publications

SpecFlow: A Specular Reflection Restoration Framework using Flow-Guided Video Completion

Prostate Lesion Detection and Salient Feature Assessment Using Zone-Based Classifiers

Selected Projects

• pcavision: Code repository for the paper “Prostate Lesion Detection and Salient Feature Assessment Using Zone-Based Classifiers” (Python/OpenCV)
• LeetCode Practice: Record of completed LeetCode questions (Python/C++)
• PyTorch Practice: Learn PyTorch with me! (PyTorch/OpenCV)
• hotcold optimize: Code repository for my reinforcement learning project at Arion in optimizing cloud storage based on user behavior (PyTorch/stable-baselines3)

Honors

• SPIE Photonics West 2023 Speaker (Accepted)
• Cornelius Vanderbilt Scholar (2021 - Present)
• Equitable Excellence $10k Scholarship (2021)
• National Merit Scholar (2021)
• Coca-Cola Scholarship Semifinalist (2020)
• Science Olympiad National Medalist (2019-2021)
• USA Biology Olympiad Semifinalist (2018-2020)

Technical Skills

• Programming Languages: Proficient in Python and C++. Knowledge of MATLAB, Java, JavaScript, SQL
• Machine/Deep Learning: PyTorch, OpenCV, Pandas, NumPy, Matplotlib, Optuna
• Web Development: HTML, CSS, Gradio, HuggingFace
• DevOps: Hydra, Git, Linux Ubuntu, Weights & Biases

Leadership

• Take detailed notes during weekly planning meetings and brainstorm organizational structure and projects
• Facilitated marketing campaign for a week to attract 100+ students to attend first workshop

• Coordinate cold calls and manage existing sponsor relations to raise $80,000 as funding for Vanderbilt's fall hackathon event
• Represent VandyHacks organization at conferences and networking events to source 40% more leads

• Scheduled viola section practices in addition to regular orchestra practice sessions
• Provided encouragement to slower members and coached them in quality improvement
• Kept section in time with orchestra and efficiently incorporated notes from the conductor

• Coordinated four-figure sponsorship from top 15 (by market cap) blockchain company
• Represented organization at Consensus Austin conference to source additional funding and speakers

• Sponsorship Coordinator @ VandyHacks, Feb 2022 - Present
• Viola Section Leader @ Vanderbilt Commodore Orchestra, Sept 2021 - Present