AI assistent

[neviaumi]

I’m considering the idea of feeding portfolio HTML into an AI model and prompting from there.

The project resume.json already explores a similar approach—it takes a resume and a job description as input and auto-generates interview-related content.

If I can fine-tune and provide access to the model, I could:

• Reduce response time
• Improve semantic accuracy since HTML is self-contained and well-structured
• Potentially generate a complete resume directly from the model
• Integrate features like an FAQ assistant, JD match helper, etc.

However, cost is a concern. How much would I need to pay for ChatGPT usage? If ChatGPT is too expensive, what are some alternatives?

[neviaumi]

The release flow can be predictable as

• Release the slower version of chatbot (It fetch the html everytime)
• Release the version with vector database in front of prompts

This i can get it work and improve later