Inez Tandy Ortega
All workAI / Data

AI-Powered
Feedback System

An evidence-informed assessment feedback system for educators and students.

MSc Computer Science / Dissertation

Development
Python · Streamlit · Pandas · NumPy · SQLite · GPT-4 API
Research
Evidence-informed redesign and mixed-methods evaluation

Assessment data.
Actionable feedback.

The Enhanced Feedback System turns assessment results into cohort analysis, individual performance views and structured feedback. It builds on an earlier system, with changes guided by a preliminary evaluation with six participants.

Educators can generate, edit, moderate and download feedback. A Teaching Summary brings cohort priorities and individual needs into one view, while shared generation logic keeps the feedback structure consistent.

Watch the actual application

Narrated demonstration / 14:30
Full demonstration · 14:30

Watch the full narrated recording, or jump to a section of the real application.

Dissertation presentation

11 slides / Complete presentation
1 / 11
Read the slide text
Evaluating the Perceived Effectiveness of an AI-Powered Assessment Feedback System in Further and Higher Education Inez Tandy Ortega MSc Computer Science Dissertation

The feedback system explained

3 slides / Complete presentation
1 / 3
Read the slide text
First, upload assessment data files. Analyses the whole cohort and categorises students into named groups based on their overall mark. Shows how each group performed across individual questions. Displays an individual student’s performance on the assessment and compares it with the cohort average. Displays statistics, including standard deviation, for cohort performance on each question. Provides actionable feedback notes for educators on overall cohort performance showing strongest and weakest topics, priority teaching areas, and the weakest-scoring questions. Also presents an overview of individual students’ strongest and weakest topics, with cohort comparisons. Generates editable feedback for students and a Feedback Management System to track whose feedback has been generated, edited, moderated, and downloaded. Presents graphs that compare last year’s cohort performance with the current cohort’s performance. Provides access to the uploaded data files. PAGE NAVIGATION FOR THE AI-POWERED FEEDBACK SYSTEM

The application journey

One assessment.
Connected views.

  1. Upload assessment dataLoad Learn, Gradescope or other assessment files.
  2. Cohort StatisticsExplore overall marks, performance groups and question-level results.
  3. Student StatisticsCompare individual results with cohort averages.
  4. Question AnalysisExamine question statistics, including standard deviation.
  5. Teaching SummaryReview strengths, weaker topics, priority questions and individual needs.
  6. Feedback GeneratorGenerate structured feedback with optional resources and practice questions.
  7. Comparison Analysis & Data SourcesCompare cohorts and access the uploaded files.

Educator control

Generate. Edit.
Moderate. Export.

The SQLite-backed workflow stores original and edited feedback separately. Editing or regenerating feedback resets outdated moderation and export states, so educator review remains part of the release process.

Evaluation / Perceived effectiveness

What participants
thought.

74Participants
81.1%Overall satisfaction · 60 of 74
37 + 37HE students + educators

Participants valued structured feedback, educator editing and the Teaching Summary. Trust depended on traceability and human oversight. The study evaluated perceived effectiveness, rather than live institutional time savings or improved learning outcomes.

Research presentationSystem overview
Return to the work
Artwork detail