Keeper.ai

Keeper.ai Job Tool
Keeper.ai AI Dictation Assistant

Inspiration - Our families have faced issues regarding travel nursing accommodations for our grandparents, whether it be in the United States or overseas. After watching our grandparents deal with inconveniences that poor travel nursing onboarding systems currently have, our fear for our loved ones prompted us to pursue this topic in this year's iteration of CalHacks. Our inspiration further grew when we did some customer discovery of some travel nurses that we could connect with. They told us about the various financial problems and constraints that they faced due to recruiting agencies taking advantage of them. Hearing these nurses talk about their "hair-on-fire" problems in onboarding, job searching, and budgeting, also motivated us to pursue this area.

What it does

Keeper.ai is an AI tool that brings the travel nursing onboarding and management space into the new tech era. Travel Nurses currently suffer from being underpaid by nursing recruitment agencies and these are nurses are desperately looking for a way out to be on their own. Keeper offers that exit. Beyond their job search, Keeper also offers tools to help patients adjust to the rapid onboarding processes by providing instant data feedback via a simple AI dictation system. Finally, Keeper acknowledges the various issues that travel nurses face with their taxes, namely naming their "home state", and how to minimize the risk of "double paying" taxes via a linear regression model.

How we built it

We built the AI job search tool by scraping a large website of 100,000 jobs and storing them in a database, in which we then filtered this database based on user input of their desired state of choice, role, or employment type. After this, we outputted 3-6 relevant jobs that fit their search by passing this input into the ever-so-fast groq API, in which they could then choose to apply and/or see their potential tax implications. Our tax implications model was built using a linear regression model to predict spending over the duration of the contract, as well as other relevant factors like state income tax, current taxable income, etc. The icing on the cake for this feature lies in Keeper outputting a completed tax report for our user. This tax feature is also coupled with an AI assistant courtesy of You.com, which provides specific legal advice on tax regulations specifically for travel nurses, so this assistant can answer any relevant travel-nurse tax related question. This was accomplished using a RAG model via langchain and pinecone. Finally, our final feature is an AI dictation assistant. Travel Nurses can start a voice recording over the time period they help patients. This assistant takes note of the entire conversation, and then the minute the travel nurse says "terminate", the dictation would stop immediately and a pdf of a relevant patient summary would be created for the nurse to share with the patient. This was done using OpenAI's speech-to-text model, as well as some simple python PDF libraries to scrape and write to.

Challenges we ran into

Frontend Dev - we were short on our team for a front end dev. We wish that we had reached out to someone at this hackathon who was skilled in the front end development, as we were mainly skilled in backend development. We struggled on that part during this hacakthon.

Accomplishments that we're proud of

This is our first hackathon, so it was really nice to see an efficient use of API's. It was also really cool to see our backend scripts working, especially with the speech-to-text model and our linear regression model.

What we learned

We learned that while figuring out the coding backend is important, so is managing your time, and keeping your code organized when working and gearing up for production.