Calvin Laughlin

Hello! My name is Calvin Laughlin.

I'm from Los Angeles, California, and I like to make things for people using a computer. I hold a Bachelor's in Computer Science from Stanford University, and I'm currently getting my Master's (also) in Computer Science.

Throughout my journey in the field, I've made a lot of things, such as a machine learning model that predicts surf, a music video generator (seen above), and a Stanford parody video game.

Experience

Tech Lead

Led a team to develop a goal-achievement web app for college students with ADHD. Tested a variety of LLMs such as LLaMA, Mistral, and GPT, designed, and implemented a novel solution to solve the company’s problem, and developed the MVP featuring user login, file upload with LLM summarization, Google calendar integration, and an A.I. chatbot. Currently being tested by 10 college students.

Section Leader, CS106A

Guided 12 students weekly through Stanford's CS106A course, clarifying complex concepts and reinforcing key materials. Evaluated 336 assignments and over 600 exam questions, conducted 24 hours of interactive grading sessions, resulting in an A average student grade. Earned an outstanding rating of 4.84 out of 5, reflecting exceptional teaching effectiveness and a profound impact on student learning.

My Resume

Projects

Large Language Lego Models

Fine-tuned LLaMa-2-7B, Mistral-7B, and GPT-3.5-turbo on text-based LEGO instructions with the goal of generating novel instructions. Cleaned data from Bricks for the Blind using Python, and translated visual instructions to text using GPT-4o vision.

Music2Dance

Developed a feature-based audio mosaic tool that aligns music spectrums with iconic dance video clips using K-Nearest Neighbors (KNN) for a coherent audiovisual performance. Audio programmed with ChucK and visuals connected using Open Sound Control (OSC).

Open-Surfline

Predicted wave heights in Monterey Bay from time series data using a variety of methods such as linear regression, XGBoost, multi-layer perceptron, and long short-term memory (LSTM). Cleaned data from the National Data Buoy Center and used Python and PyTorch to make predictions.

Hack OS

Terminal-based puzzle story game that lets players feel like they're hacking without knowing how to hack. Coded in Python with the Curses library for CS247G: Design for Play.

Quad Contest

Top-down 2D roguelike that takes place in Stanford Main Quad. Built in Unity and coded in C# for independent study directed by Jay Borenstein. Defend yourself against waves of bears as they invade Stanford!

Wekinators

Throwing my computer up in the air like a child, controlling music tempo using Wikipedia tabs, and playing Minecraft with my face and my voice.

Cal-Train

Original personal website that simulates waiting for the CalTrain at the Palo Alto station. Walk over to the ticket machine and press space for a surprise.
Built with Javascript and HTML, optimized for Desktop.

Beat-Boxing Imitation Game

Not the kind where you try to distinguish human from machine, but instead kind where you try to fight for your humanity by replicating the commands given to you by the computer.

Sonic Poetry

Experimental terminal poetry using the audio programming language ChucK and chAI (ChucK for AI), with the AI being utilized through Word2Vec to create machine-generated words within the poems.

Hello! My name is Calvin Laughlin.

I'm from Los Angeles, California, and I like to make things for people using a computer. I hold a Bachelor's in Computer Science from Stanford University, and I'm currently getting my Master's (also) in Computer Science.

Throughout my journey in the field, I've made a lot of things, such as a machine learning model that predicts surf, a music video generator (seen above), and a Stanford parody video game.

Tech Lead

Section Leader, CS106A

Large Language Lego Models

Fine-tuned LLaMa-2-7B, Mistral-7B, and GPT-3.5-turbo on text-based LEGO instructions with the goal of generating novel instructions. Cleaned data from Bricks for the Blind using Python, and translated visual instructions to text using GPT-4o vision.

Music2Dance

Developed a feature-based audio mosaic tool that aligns music spectrums with iconic dance video clips using K-Nearest Neighbors (KNN) for a coherent audiovisual performance. Audio programmed with ChucK and visuals connected using Open Sound Control (OSC).

Open-Surfline

Predicted wave heights in Monterey Bay from time series data using a variety of methods such as linear regression, XGBoost, multi-layer perceptron, and long short-term memory (LSTM). Cleaned data from the National Data Buoy Center and used Python and PyTorch to make predictions.

Hack OS

Terminal-based puzzle story game that lets players feel like they're hacking without knowing how to hack. Coded in Python with the Curses library for CS247G: Design for Play.

Quad Contest

Top-down 2D roguelike that takes place in Stanford Main Quad. Built in Unity and coded in C# for independent study directed by Jay Borenstein. Defend yourself against waves of bears as they invade Stanford!

Wekinators

Throwing my computer up in the air like a child, controlling music tempo using Wikipedia tabs, and playing Minecraft with my face and my voice.

Cal-Train

Original personal website that simulates waiting for the CalTrain at the Palo Alto station. Walk over to the ticket machine and press space for a surprise.Built with Javascript and HTML, optimized for Desktop.

Beat-Boxing Imitation Game

Not the kind where you try to distinguish human from machine, but instead kind where you try to fight for your humanity by replicating the commands given to you by the computer.

Sonic Poetry

Experimental terminal poetry using the audio programming language ChucK and chAI (ChucK for AI), with the AI being utilized through Word2Vec to create machine-generated words within the poems.

Original personal website that simulates waiting for the CalTrain at the Palo Alto station. Walk over to the ticket machine and press space for a surprise.
Built with Javascript and HTML, optimized for Desktop.