Selected work
Systems I’ve
built and studied.
A small archive of projects that show how I think across product engineering, machine learning, and systems.
Discord Automation Flows
Syndicated data from internal databases and the web, creating a daily summary of postings for internship seekers.
What I built
Joined multiple calendars into one through cross-platform automation, creating a singular view.
How it grew
Built and maintained a Discord bot with Gemini and Notion integrations serving over 1,000+ active users, with commands for event syncing, message reactions, and internship notifications.
ClearSky - Unwrapathon
Worked as a team of 5 to develop a data pipeline that visualized user sentiment across subreddits.
The team
Built the project as part of a five-person team focused on visualizing sentiment across subreddits.
My work
Wrote scripts to fetch and clean data returned by the Reddit API and perform sentiment analysis.
Physics Engine Evaluation
Conducted performance testing on local devices, AWS EC2, and Brev to compare Genesis and Isaac Gym.
What I tested
Evaluated 5 unique simulations running in up to 200,000 environments simultaneously, rendering millions of FPS.
What I built
Automated testing and data visualization with Python scripts, streamlining tests and highlighting key metrics.
TinyFS and Disk Emulator
Implemented a disk emulator that handled basic Unix file block operations: read, write, open, and close.
The file system
Created a TinyFS file system handling basic file operations, timestamps, fragmentation, file permissions, directory listing, and file renaming.
Testing
Used visualization software to run automated and manual testing with custom C++ scripts.
Mustang Chat
Developed a multi-user, client-server chat application by implementing a custom network protocol over UDP.
Channels
Implemented server-side logic for dynamic channel management, including implicit creation on first join.
Reliability
Designed and implemented soft state session management to ensure reliability over connectionless UDP.
Data Analytics and Reporting Pipeline
Developed and tested SQL-based data pipelines and integrations across multiple relational datasets.
Data quality
Validated data flow, transformation logic, and edge cases while identifying and documenting inconsistencies and errors.
Automation
Automated reporting and analytical workflows, reducing manual effort and improving data accuracy by 10%, with deduplication, NULL handling, and consistency checks.
Sentiment Classifier
Built a sentiment classifier to derive meaning from user comments on videos using neural networks with PyTorch.
Modeling
Conducted grid search to compare various neural network architectures, including RNN, GRU, and LSTM.
Focus
Used Python, PyTorch, and Scikit-learn to evaluate neural-network approaches for comment sentiment.
Find Your Niche
Converted the listening history of 54 users containing 7 million song plays into clean dense matrices for analysis.
Clustering
Categorized users into musical niches based on density-based scanning and k-means clustering.
Recommendations
Used collaborative filtering and KNN to recommend new artists and songs based on user similarity.