PROFE : Personalized Resources Optimized For Education

Profe AI Agents Architecture

Inspiration

PROFE emerges from the critical need to address high dropout rates and the difficulties students encounter in completing their school assignments and grasping complex subjects. In Latin America, traditional tutoring services are often financially out of reach, and educational materials in Spanish are limited. Our mission is to democratize access to educational support, providing an affordable and personalized solution for every student.

We envision a learning environment that is inclusive and accessible, breaking down the barriers that hinder students' academic success. With PROFE, students have the opportunity to access a personalized tutor tailored to their specific needs. They can share their own class notes and study materials, creating a truly customized learning experience.

Furthermore, the availability of PROFE 24/7 ensures continuous support, and the option to interact via WhatsApp makes communication easy and convenient. Our goal with PROFE is to eliminate traditional barriers to tutoring access and create a responsive learning environment that caters to the unique needs of every student, regardless of their circumstances.

What it does

PROFE provides an interactive teacher powered by Gemini large language models, accessible via WhatsApp. It assists students in understanding various topics from school, allowing them to share images or PDFs of their class notes and ask questions. PROFE can generate sample quizzes based on any topic provided or from the student's class notes, and can respond with text or audio.

How we built it

We built PROFE using Python and FastAPI. We used the ReAct LLM agent paradigm implemented using Langchain, we utilize Gemini large language models for text and vision processing, as well as Google Text-to-Speech and Speech-to-Text APIs.

PROFE consists of an orchestrator agent that receives the student messages and decides which agent to call to construct the answer to the student. The available agents are:

Enrichment Agent: Enrich student information with data from Wikipedia.
Interpretation Agent: Helps student to interpret images and PDF files.
Quiz Agent: Generates quizzes based on the students requests.

PROFE facilitates communication with visually impaired students by interpreting audio inputs and delivering responses in audio format.

Additionally, to integrate the API with WhatsApp, we used Twilio and ngrok for local development.

Challenges we ran into

Integrating the Gemini API and ensuring proper authentication
Managing dependencies and environment setup across different platforms
Addressing compatibility issues with different versions of Python and third-party libraries

Accomplishments that we're proud of

Integrating natural language processing capabilities to understand user queries effectively
Developing a user-friendly interface accessible via WhatsApp
Overcoming technical challenges and ensuring smooth communication between different components of the system

What we learned

Enhanced understanding of natural language processing techniques and their practical applications
Improved proficiency in Python programming and working with APIs
Gain insights into the challenges and opportunities in developing educational technology solutions
Enhanced collaboration and teamwork skills through project development and coordination

What's next for PROFE

Moving forward, we plan to enhance the personalization features of PROFE to better adapt to individual learning preferences and styles. This includes refining the AI's responsiveness to the unique needs of introverted students and those who require extra support outside conventional learning hours. Our ultimate goal is to expand PROFE's capabilities to not only assist with academic challenges but also to inspire and motivate students to achieve their fullest potential.