[{"content":"The Problem Grading 100+ IT practical assessments by hand takes hours — checking Delphi code for correctness, verifying SQL queries produce the right output, and evaluating theory answers against CAPS rubrics. With three grades (10, 11, 12) and multiple classes, the marking load was unsustainable.\nThe Solution: MaaS MaaS (Marking-as-a-Service) is a multi-agent pipeline that:\nParses student submissions (Delphi .pas files, SQL scripts, and theory answer sheets) Evaluates each component against …","date":"2026-06-25","permalink":"/posts/automated-marking-pipeline/","summary":"How I built MaaS (Marking-as-a-Service) to automate the grading of IT practical and theory assessments using multi-agent AI orchestration.","tags":["ai","education","automation","hermes-agent"],"title":"Building an Automated Marking Pipeline with AI Agents"}]