Comment by igor47

I'm working on a DnD character sheet app! I spent last week implementing the core DnD SRD ruleset, but what I'm really excited about is ML integration. I want to add a self-hosted fine-tuned ML model that acts as a character and DM assistant. Obviously an LLM via API can do the job, but I'm really curious if it's possible to build smaller, cheaper, task-specific models. Plus, I've never integrated an ML model into a product before, and I'm curious to play with it. I'm thinking of it like clippy for DnD: "it looks like you're trying to cast fireball?"

Besides the LLM experimentation, this project has allowed me to dive into interesting new tech stacks. I'm working in Hono on Bun, writing server-side components in JSX and then updating the UI via htmx. I'm really happy with how it's coming together so far!