AI Pair Engineer is a lightweight AI-assisted code review tool that helps developers improve short code snippets before human review.

The application analyzes a piece of Python or JavaScript code and returns structured feedback including code quality issues, maintainability improvements, suggested test cases, and a refactored version of the code.

This project demonstrates how AI can act as a pair programming assistant, helping developers identify potential problems early and improve code quality before submitting changes for review.

Code Quality Review Detects readability, structure, and maintainability issues in a snippet.

Three Targeted Improvements Returns exactly three meaningful suggestions to improve the code.

UX Awareness Highlights potential user-experience improvements when the code affects user interaction.

Suggested Test Cases Recommends practical tests developers should implement.

Technical Positive Note Includes one professional observation written like a senior engineer reviewing the code.

Refactored Version Generates a cleaner version of the submitted code.

Language Awareness The review adapts to the selected programming language (Python or JavaScript).

Fallback Review Mode If the OpenAI API is unavailable, the tool still provides a basic local review.

Paste a short function or component into the interface and click Review Code to receive structured feedback.

Example snippet:

Example feedback:

Streamlit – interactive UI for the tool

OpenAI API – AI code review generation

JSON structured responses – reliable parsing of AI output

Clone the repository:

Create a virtual environment:

Activate the environment:

Install dependencies:

Create an API key from:

Set the environment variable.

Then run the app:

Code review is one of the most valuable but time-consuming parts of software development.

This prototype explores how an AI assistant can act as a pre-review step, helping developers:

catch issues earlier

improve readability and maintainability

think about test coverage

produce cleaner code before human review

The goal is not to replace human review, but to augment developer workflows.

Potential enhancements for this prototype include:

Code quality scoring system

Severity levels for detected issues

Before vs after code comparison

Support for additional programming languages

GitHub pull request integration

Automated test generation

Sara Mahran Software Engineer | Coding Instructor

GitHub: https://github.com/SaraMahran