Projects

ClawdBot is my personal self-hosted AI assistant built around OpenClaw, a gateway that connects chat apps to AI coding agents and automation tools. The repository documents the setup, configuration, troubleshooting, and long-term plan for running my own assistant on a VPS.

The bot is connected to Telegram, so I can interact with it from my phone as a normal chat interface. From there, it can help with GitHub work, read and write notes, search the web, check calendar and task information, and run scheduled workflows.

One major use case is a centralized notes system: Obsidian is the main editor on my Mac, GitHub acts as the sync layer, and ClawdBot can read, search, create, and update notes through Telegram. This lets me capture ideas, meeting notes, and project context quickly while keeping everything in a Markdown-based knowledge base.

The project also includes a daily planning system with Google Calendar, Google Tasks, habit tracking, self-hosted SearXNG search, market/news lookups, and a scheduled morning briefing delivered through Telegram. I also documented team-access plans for Tapd, where different agents can have scoped access to shared notes, email, calendar, and GitHub resources without exposing my personal data.

The repository is currently private. Email me if you would like access.

** note to self: add before and after video. (on my pc)

Built a Python CLI tool to improve the quality of short-form vertical videos while learning more about Python, FFmpeg, and practical video-processing workflows.

The tool uses FFmpeg motion interpolation to target 60 FPS playback, then applies a second processing pass that scales videos to 1080x1920 and sharpens the output with an unsharp filter. Audio is copied through without re-encoding.

I built this while running two short-form sports edit accounts, one focused on basketball and one focused on MMA, where I wanted posted videos to have the highest possible quality. Over roughly two months of posting, the best-performing basketball video reached about 600K views and the best-performing MMA video reached about 200K views.

The repository is currently private. Email me if you would like access.

Alongside Computational Mechanics coursework, completed a 10-week research project with Professor Nelia Mann analyzing the elastic pendulum to construct numerical methods useful for studying nonlinear dynamical systems.

Presented the research in a poster session for the Physics and Computer Science departments, covering analyses of chaotic behavior including bifurcation diagrams and chaotic ranges of initial conditions.

Built a series of Python notebooks implementing the numerical methods used to analyze the system, including Runge-Kutta 4, linear multistep methods, Lyapunov exponent calculators, and other computational tools for exploring nonlinear motion.