EDict (Table 66)

Inspiration

We have noticed that a majority of students shy from asking questions during lectures. The reasons may vary but this feedback loop is valuable to retain information and understanding new and challenging concepts.

What it does

The goal for this application is to utilize powerful machine learning tools in conjunction with high quality sources of information directly provided by the teachers, including transcripts of lectures. When the student asks the question, the application parses everything the teacher said during the lecture along with the uploaded material and answers. In addition, the teacher will have access to all the questions asked, so if there is a question that the AI can't answer from the content, then the teacher can answer.

How we built it

We have crafted the application to be as composable as possible. All the pieces such as vector database for RAG, LLM model API, voice to text generation are separate microservices that ensure scalability of application. Using serverless functions, we can provide stronger hardware only to the pieces of our project that need it.

Challenges we ran into

The main challenge we had was delegating responsibilities since we each had various skill sets but the project required many different responsibilities.

Accomplishments that we're proud of

All of us are proud of the quality of the application that we were able to create in less than 36 hours. We are very happy with the quality of answers we were able to get and the performance of the application overall.