AI Math Tutor

Step-by-step math help in a Streamlit app, with structured outputs and optional tracing

AI Math Tutor is a Streamlit app that walks through math problems with explanations, using LangChain against an OpenAI model and Pydantic for structured, typed responses. LangSmith tracing is optional; the README notes the core tutor still runs if you skip it. Sample prompts help you try the flow quickly.

When it is useful

You want a sandbox for model-assisted study, you are demoing LangChain + structured outputs, or you need a reference layout for observability hooks. It is not a replacement for a human teacher, official coursework policy, or exam rules.

What you can do

Submit problems and read step-style breakdowns in the UI shown in the project media.
Use bundled examples to learn the flow before typing your own prompts.
Enable LangSmith when you want traces for debugging or tuning (keys and setup are in the repository).

Limits

Models can err on arithmetic or reasoning; verify critical steps, especially for assessments or safety-related math.
Cost and data handling follow your OpenAI account and what you choose to log; review terms before pasting private or proprietary material.
Depth and curriculum alignment depend on prompting and model behavior, not on this portfolio page.

Frequently asked questions

What is AI Math Tutor?

It is a Streamlit app that walks through math problems with step-by-step explanations using LangChain, an OpenAI model, and Pydantic structured outputs. Sample problems are included to try the flow quickly.

How do I run it?

Needs Python 3.11+, an OpenAI API key, and uv. uv sync, set OPENAI_API_KEY in .env (see .env.sample), then uv run streamlit run main.py at localhost:8501.

Is LangSmith required?

No. LangSmith tracing is optional. Without it, tracing is disabled but the core tutor still runs.

Can I trust every solution?

No. Models can err on arithmetic or reasoning, so verify critical steps especially for assessments. Cost and data handling follow your OpenAI account.