MA Grader

Overview

Grading spreadsheet assignments by hand means opening every file and checking dozens of formulas one cell at a time. MA Grader is for instructors teaching a college math course built around Excel assignments. The instructor drops in a ZIP of submissions and picks the assignment. The app grades every workbook, writes a graded sheet with comments for each student, and builds an instructor master sheet.

What I built

• A Python grading engine (about 14,800 lines across 122 files) that checks formulas, values, formatting and charts in student workbooks.
• A formula-equivalence checker that tests a student’s formula against the correct one using random inputs, so correct variations get credit and hard-coded answers don’t.
• An Electron + React desktop interface with drag-and-drop and live grading progress.
• Optional AI grading of the written paper against an 8-criterion rubric, using the instructor’s own API key stored encrypted on their machine.
• One-click installers. The engine is packaged so users don’t need Python, and GitHub Actions builds Windows and macOS installers for each release.
• A grading audit that runs correct and incorrect formula variants through the graders to catch grading mistakes.

Impact

• A batch of 20 major assignments used to take about 3 hours by hand (8–10 minutes each). It now takes about 30 seconds to grade, plus around 10 minutes of manual verification.
• Close to 800 major assignments graded so far, an estimated 100+ hours of grading time saved.
• Used by two instructors across our course sections.

How it was built

I designed and directed this project and wrote much of the code with an AI coding assistant (Claude).