Referency

An AI-powered Recommendation Letter Generator that automates and simplifies the recommendation letter writing process.

Comment

Inspiration

In my school, where we have a few staff but multiple students applying for different graduate school opportunities, lecturers were getting overwhelmed because they had to write many recommendation letters besides their already intense schedule.

What it does

The application allows students to enter information such as their final capstone project alongside the grade, conversations they have had with the faculty to whom they want a letter, the opportunity they are applying for, and other relevant info. The student sends a request to the faculty. On the other hand, the faculty receives an email that they have received a request and can see everything about it. The faculty can choose to generate a letter based on the requested information.

How we built it

The application was built with Flutter for the UI, Firebase for the database operation, and Google AI with the Gemini Pro model for the LLM.

Challenges we ran into

The first challenge was trying to streamline the whole letter recommendation process since I had to ensure each letter generated was unique to the student. The second challenge was deploying the letter recommendation portion of the project as a cloud function and integrating it into the Flutter application. I stumbled across the CORS error.

Accomplishments that we're proud of

I was proud to see that by collecting some relevant information from the student, an adequate letter was generated that could be edited with a few changes, which will help a student prepare a strong graduate school application.

What we learned

  1. Working with cloud functions.
  2. Crafting a prompt based on several pieces of information to make the model output something useful.
  3. Working with the Gemini Pro model and integrating it into an application.

What's next for Referency

  1. Providing an avenue for feedback to be sent back to the LLM so it can improve upon the letter.
  2. Integrating a faculty's writing style so that each letter provided by different lecturers for the same student will make a difference in the writing style.
Share this project:

Updates