tutor

AI Agents for Personalized Learning & Tutoring

Features

Feature	Status	Comment
Student modeling	✅	Multiple stupdents supported
Assess learning gaps and Assess knowledge retention	✅	Using quizzes
Multiple AI tutors for Different subjects	✅	Using the same agent, but supports multiple subjects
Different teaching styles	✅	Suport for textbook and podcast style
Agents collaborate	✅	Study guides are asynchronously updated on every quiz answer
Suggest interdisciplinary connections between topics	❌	Setup to update topic study guides, but LLM to recommend which study guides to update is not implemented
Interactive exercises	✅	Selecting text in study guide allows to ask questions about the selected text
Quizzes	✅
Gamification elements	✅	Difficulty level grows with every question answered correctly. Or it falls if question is answered incorrectly. Agent decides when and by how much to increase the level of difficulty given all the questions answered so far
Event-driven patterns	✅	Study Guide Agent sends quiz answers to --> Study Progress summarizes weaknesses and user proficiency level --> Study Guide Agent generates new study guide and quiz questions best on weaknesses and user proficiency level
Agents maintain their state across events	✅	Using in memory persistence, but maybe should use a database if time allows
Human interaction	❌	Need to add explanations on study guide and hints on quiz
Integrate with existing process systems	✅	Pulling data from database to create study guide
Ethical implications	❌	Need agents to double check errors or allow to report errors

Setup instructions

Pre-requisites

Setup Kafka and run the server

docker pull apache/kafka:4.0.0
docker run -p 9092:9092 apache/kafka:4.0.0

Copy the .env.sample to .env, and DON'T FORGET to fill it out. Currently I'm using only OpenAI models, so likely only need to fill out the OPENAI_API_KEY.

cp .env.sample .env

Running the app

pip install -r requirements.txt
python app.py   

Presentation

After clicking the link, click the download button at the top right corner. Link to the presentation Link to the demo

Architecture

Obsolete Google Cloud instructions (Since I'm not using Google Cloud Services anymore should not be needed)

gcloud config set project <PROJECT_ID>

export PROJECT_ID=$(gcloud config get project)
export SERVICE_ACCOUNT_NAME=$(gcloud compute project-info describe --format="value(defaultServiceAccount)")

gcloud services enable compute.googleapis.com  \
                        aiplatform.googleapis.com \
                        secretmanager.googleapis.com \
                        cloudfunctions.googleapis.com

Setup Permissions:

gcloud projects add-iam-policy-binding $PROJECT_ID \
  --member="serviceAccount:$SERVICE_ACCOUNT_NAME" \
  --role="roles/aiplatform.user"

gcloud projects add-iam-policy-binding $PROJECT_ID \
  --member="serviceAccount:$SERVICE_ACCOUNT_NAME" \
  --role="roles/secretmanager.secretAccessor"

Generate a service account key:

gcloud iam service-accounts keys create key.json \
  --iam-account=$SERVICE_ACCOUNT_NAME