Campus Feedback Intelligence
A privacy-preserving pipeline that clusters free-text student feedback into recurring themes and tracks them over time.
The problem
Feedback arrives in volume and gets reduced to an average. The specific, repeated, actionable thing that forty people wrote is exactly the information the average destroys.
What we are building
A clustering pipeline that groups comments by subject rather than sentiment.
Anonymisation and minimum-cluster-size rules enforced in the pipeline, not in the interface.
A longitudinal view showing whether a theme is growing or fading.
Current status
The most useful thing a project page can do is be specific about what is not true yet.
- Exploratory work on clustering approaches using synthetic data.
- A first draft of the privacy rules.
- No real feedback data is being used, and none will be without consent and a review.
- No validated methodology yet — cluster quality is still being checked manually.
Technology
Team
Position open
Position open
Open roles
Pick the one that sounds like the thing you want to get better at.
- DataClustering pipeline and evaluation of cluster quality
- ResearchPrivacy guarantees and methodology
- FrontendAnalysis interface and longitudinal views
What you can contribute
Hand-label a sample so the clustering has something to be measured against.
Write the privacy rules as executable tests.
Design a visualisation that shows change over time without implying causation.
Timeline
Define
Complete
Problem framed; privacy treated as a hard constraint.
Research
In progress
Clustering approaches evaluated on synthetic data.
Build
Next
Pipeline with anonymisation enforced at ingestion.
Analysis
Later
Longitudinal theme tracking.
Want in?
Apply to a role, or start by reading the problem behind it. Neither commits you to anything.