Abstract
This isn't just another chatbot. It's my digital persona, brought to life through AI. I've carefully engineered its core identity to think, joke, and respond the way I do. Ask it about my projects, my opinions on tech, or just have a casual conversation. It's a living experiment in identity, AI, and the art of instruction—all through the power of sophisticated prompt engineering.
Process
I implementated the Open API model gpt-3.5-turbo for the chatbot. Then I fed JSON data of my personality and likeness to the bot. Thus, this chatbot uses data from both the gpt model JSON file as it's corpus. The logic to retrieve the response from the data is in the express.js file where it is constantly run in the hosted cloud server: render.com. The UI and frontend is also being constantly run in another render.com server. The frontend has pre made prompts for the user to enter.
Reflection
Unfortunately, the MongoDB database deployment was unable to capture the weighted scoring for the prompts and could not capture my customized data to full effect. So my data is currently localized in the backend server. I'm currently working on finding a solution and trying to find NoSQL alternatives.