TC On the Line

Personal Projects

Here are some of my personal projects.

Project 1: This website

This website was inspired by 90's web design. and I had three main goals in mind when I started it:

  1. Learn about web design standards from the 90's. (Specifically 1997)
  2. Make the website extremely efficient and fast to load. (Around 50KB of data per page)
  3. Keep the aesthetic of the 90's while making it modern and responsive.

These goals led me to a lot of research and learning that I would never had done if I had not set these goals for myself. I have a page here you can view that shows some of the learnings I had during this project.

View Learnings

Project 2: LLMFunWebsite

LLMFunWebsite was a project I started to get used to calling LLMs to create different workflows. I chose to do a website because I had the most experience with it and liked the idea of being able to share easily. The primary goals were to:

  1. Learn LLM libraries and how to use them. (OpenAI's API and Langchain)
  2. Create a secure website and backend that would not allow for my LLM costs to get out of control.
  3. Have a good sandbox to play in and learn from.

I ended up learning a lot about Langchain and the common API patterns for using LLMs. I had a couple projects that I worked on that were part of that sandbox that never truly got finished, but I still learned a lot from them.

Project 3: Rap Battle

I had always wanted to make a social project and had been trying to come up with ideas. This idea came from friends of mine that had been making music and coming up with lyrics. I've thought songwriting is really cool and difficult so I wanted to make something we could enjoy and practice with. The idea was to create a website that would group users to compete in lyric writing challenges. Two users from the group would be chosen to create a simple 4 line verse and the rest of the group would vote on the best verse.

Another primary reason was to get more experience with agentic coding and LLM workflows. I wanted to see how far I could push the agent without my own intervention. Turns out, it was pretty far.