Logan Yuan

Leading a team of 4 research assistants to update GailBot, an audio transcription service used by 100+ researchers at universities across the United States.

Managing project scope, timelines, and resources for 100% on-time project completion.

Developed a React-based platform for researchers to share and manage machine learning transcription software.

Designed backend architecture, designing and implementing database tables and fields. Used AWS (MySQL RDS and S3)

Coded plugins for AI analysis of audio files using PyTorch

Analyzed drilling efficiency of shale formations using deep learning and PyTorch.

Built and trained ML models, such as convolutional neural networks, linear regression, trees, etc.

Handled entire training pipeline, including preprocessing, feature extraction, training, and evaluation.

Built a user feedback system for a local homelessness nonprofit with 1000+ clients.

Coded REST APIs to a MongoDB database in Node.js and Prisma.

Used OpenCV to detect faces in an image and highlight them.

Imported the haarcascade algorithm and passed it png files for testing.

Built a Discord bot in Python to detect and redirect links sent in a text channel.

Developed a reverse Polish notation calculator in a low-level assembly emulator.

Tracked call and value stacks with pointers, implemented loops and conditionals with jump statements, managed registers for computation.

Coded an emulator for writing and executing assembly-like instructions.

Emulated CPU registers, requesting/accessing heap memory, and memory addressing.

Built an image compression/decompression tool in C.

Used quantization and bit manipulation to compress image files by 60%.

Built an image rotation tool in C.

Implemented row, column, and block traversal algorithms to explore cache optimizations.

Built and trained an automatic speech recognition model from scratch using PyTorch.

Used torchaudio to convert audio samples to Mel spectrograms, then trained CNNs on them to detect spoken features.

Trained model in the cloud by renting GPUs

Built and trained a transformer model from scratch using PyTorch.

Implemented self-attention layers, dense layers. Optimized training with stochastic gradient descent, mixed precision, and learning rate decay.

Trained model in the cloud by renting GPUs.

Coded a Grep clone in C++.

Designed a HashMap from scratch in C++ for strings, including load factor, collision handling, resizing.

GPA: 3.91

Majors: Computer Science and Mathematics

Logistics coordinator for JumboHack, Tufts main hackathon (2024-2025)

Software developer for JumboCode (2023-2024)

IT Assistant for Hartford Hospital (2023)

Teaching Assistant for AP Calc BC (2022-2023)

Pianist: Concerts, competitions, chamber, etc (2010-2023)

CS160: Algorithms

CS138: Reinforcement Learning

CS40: Machine-level Structure and Assembly

CS15: Data Structures

CS11: Introduction to Computer Science

MATH135: Real Analysis

MATH72: Linear Algebra

MATH65: Bridge to Higher Math

MATH42: Calc III