The idea behind ResumeChat was simple (or so I thought, at first.) It would target users who were looking for work. A user could upload his/her resume, and answer a few simple questions about their preferences, like “Are you willing to relocate?” and “What job titles are you looking for?” and “Do you prefer remote, hybrid, or on-site work?”
The resume and the answers to those questions are then converted to vectors and stored in a vector database. The user is given a personalized link to his/her own chatbot. He or she can then pass that link along to recruiters and hiring managers. Those people can then pull up the chatbot and ask it questions – “Is John Doe interested in contract work?” and “Is Jane willing to relocate for work?” The idea was to eliminate having to schedule the first “introductory” call between a recruiter and a candidate, and the recruiter could get a feel for the candidate at any time of the day or night, from anywhere in the world.
The entire thing took me almost six months to get up and running. I was also able to tie into Stripe’s payment API (the idea was to charge a small monthly amount for users.) Everything runs on a Google Cloud Compute instance. (Funnily enough, getting the payment API to work was almost more complicated than the LLM/RAG/GenAI stuff.)
Unfortunately, there was absolutely no interest in the application, so I took it down and shut down the GCC instance, but the code is in my github for anyone to peruse: https://github.com/wdonat/ResumeChat.