Apollo

Inspiration

Watching my son grapple with dysgraphia—a neurological condition that hinders his ability to transcribe thoughts into written words—has been both heart-wrenching and inspiring. Despite his clear understanding of the material and thorough preparation, he often struggles to complete his exams on time. To improve, he's been using practice questions and timers to quicken his response rate. However, he frequently runs into roadblocks due to a shortage of available practice material and uncertainty about the accuracy of his solutions.

Moved by his perseverance and as a programmer myself, I felt compelled to bridge this gap. Thus, I created Apollo.

What it Does

Apollo isn't just a tool; it's a beacon of hope for students like my son. This innovative exam preparation solution not only generates an endless stream of questions but also provides detailed, step-by-step answers to each one. Apollo aims to empower students, offering them the confidence and the resources they need to excel, regardless of their challenges.

How we built it

• We created an api using Gemini Pro for generating questions, generating answers and correcting a student's submitted answer
• Go is used for the microservices and a client app was created using NuxtJS

Challenges we ran into

• Proofs using trig identities would sometimes generate unsolvable proofs. To mitigate this we came up with a rubric for the LLM to use when solving these types of questions

Accomplishments that we're proud of

• My son was able to use the app to generate questions and he did indicate it helped improve the time it takes for him to solve those problems

What we learned

• It was great to revisit Vertex and checkout Gemini. I like there is more information in the response.The multi modal model is interesting as well
• There is a lot of research being done with regards to LLM's ability to solve math and generate questions.

What's next for Apollo

Built With

• go
• nuxtjs