I study Electronic Systems Engineering at Texas A&M with minors in Computer Science and Cybersecurity. I build software solutions for the problems I run into in daily life and make them usable for other people to benefit from.
Daily SAT/ACT vocabulary practice with shared progress across web and iOS.
I built the React web app, Expo iOS app, and FastAPI/MySQL backend around a spaced-repetition feed of word, quiz, and sentence cards. The API runs on Cloud Run, and both clients use the same account and progress data so a study session can continue on either device.
Clear and easy to understand.
A macOS music player for local libraries, audio tools, and a guest-controlled queue.
I built the SwiftUI player, AVFoundation playback layer, local library and playlist tools, responsive visualizers, and a party mode that lets guests queue tracks from a browser. A bundled FastAPI service and FFmpeg handle media work; track files and private paths stay on the host Mac.
Work in progress: a local research queue with saved sources and run history.
I am building Pallie because research tasks tend to scatter across browser tabs, chat histories, and loose notes. Its FastAPI/SQLite service records tasks, runs, events, and artifacts; a Codex plugin handles browser-backed research, and a SwiftUI macOS app reads the inbox and saved reports.
The worker currently completes research-only jobs and writes Markdown reports. Purchasing, booking, sending, deleting, and account changes are not implemented.
Review Pallie's job-watch pipeline and identify what blocks a reliable daily run.
Deep Trade Algo started as a way for me to understand where stock-signal demos can introduce look-ahead and evaluation bias. I built a Flask dashboard and CLI that run MA/RSI/MACD rules or a compact PyTorch LSTM, then compare a long-only simulation with buy-and-hold. Signals execute on the next row, model training and evaluation are chronological, and deterministic fallback data keeps the demo and tests reproducible.
Source codeOpen full demoMermaid Studio came from wanting a quick local editor that did not send diagram source to a hosted renderer. I built the React/TypeScript app with live Mermaid rendering, history, high-resolution PNG/SVG/source exports, a loopback artifact API, and an MCP server. Manual positioning is limited to flowcharts, where edited nodes and connectors can be preserved reliably.
Source codeOpen full demoThis party-game prototype gave me a concrete way to work through server-owned multiplayer state. I implemented room codes, hidden roles, task turns, sabotage, and voting with Flask, signed sessions, and a bounded in-memory room store. It is deliberately scoped to one computer or a trusted local network rather than presented as production multiplayer.
Source codeOpen full demoI used a saved set of 2,429 Mercedes-Benz listings to build the full path from data cleanup to a browser estimate. The Flask app trains a scikit-learn Random Forest; the static demo loads a smaller exported Ridge model from versioned JSON. A grouped holdout keeps identical model inputs on one side of the split, and the page reports model limits beside the estimate.
Source codeOpen full demoI wanted a planner that could turn a rough task list into a usable schedule without needing an API key for every run. I built a Flask app that can use OpenAI and keeps the five most recent schedules in SQLite; the static demo parses tasks deterministically in the browser and saves its latest plan in localStorage. That fallback keeps the public demo testable and independent of a paid service.
Source codeOpen full demo