ExamMaster

In the final year of my Computer Science degree, module material lived across lecture slides, notes, recordings and course resources. Preparing to revise meant switching between sources, reorganising content by hand and hunting for enough practice material to actually test understanding.

I built ExamMaster for my own revision first, then shared it with classmates. It became the revision hub for more than 30 peers during the exam period.

The framing I worked to was: how might we reduce the friction between course content and active exam practice, so students spend more time learning and less time organising revision material?

That broke down into five separate problems, only one of which was about storing content.

• Fragmentation — material was distributed across multiple files and learning systems.
• Passive revision — reading slides did not produce enough active recall or practice.
• Progress visibility — students could not see what they had covered and what remained.
• Exam readiness — students needed practice resembling the demands of testing, not a content repository.
• Time pressure — final-year students needed something immediately useful, not another platform to configure.

The first user was me. The product only became interesting when people who had not built it started using it voluntarily.

• Primary — final-year Computer Science classmates preparing for exams, and myself as the original user.
• Potential future users — students on other modules and courses, and learners preparing for professional certifications or interviews.
• Stakeholders — students whose module material is represented in the app, and the course context. ExamMaster was a student-created revision aid, not an official university learning platform.

This was founder-as-user discovery rather than a theoretical persona exercise. I experienced the friction directly, built for it, then let peer use tell me whether the problem was mine alone or the cohort's.

Every expansion after the first module was driven by watching how classmates used it and what they asked for next.

The hardest constraint was timing: a revision product is only valuable during a narrow window, and it competes with the exams it is meant to support.

Most of these decisions traded feature richness for speed of orientation, because that is what the user actually lacked.

One module, done properly, released to a handful of students. Coverage only expanded once that module was being used without prompting.

I started with my own revision workflow, defined the module structure, learning flow, question and practice experience and the progress model, then used AI-assisted development tools to accelerate implementation.

I tested the product myself, shared it with classmates and iterated around what made revision faster and easier to navigate. AI also helped draft practice questions at volume, but questions were checked against module material before reaching another student — in a learning product, an incorrect answer costs more credibility than a missing feature.

The product problem, information architecture, learning flow, content organisation, prioritisation, testing and iteration were mine.

The interface story is short by design: choose a module, review the structured material, answer practice questions, see progress, take a practice run, then return to the weak areas.

Module cards carried the practice volume and week coverage on the face of the card, so a student could judge scope before committing attention.

• Distributed Programming — 200 questions, 10 weeks.
• Cloud Computing — 200 questions, 10 weeks.
• Bloomberg Market Concepts — 160 questions, 8 weeks.
• Product Management — 240 questions, 12 weeks.
• These are documented examples from the captured version of the product; the live build may since have changed.

More than 30 Computer Science classmates used ExamMaster heavily during final-year exam preparation. Adoption was voluntary and spread by peer recommendation, which is the signal I trust most here.

Feedback was strong and qualitative: usefulness, clarity and how much it helped structure revision. Peer use proved the problem extended beyond my own workflow, and the product held up across several modules rather than a single-course prototype.

What I cannot claim: any causal link to grades, pass rates or exam performance. That would need a study I did not run.

What this project demanded of me as a product person, rather than what the product contains.

A problem I experience personally is a useful starting point, but other users decide whether it is a product opportunity.

Students under time pressure value fast orientation and continuity far more than feature richness.

Active practice is what turns a content repository into a learning workflow, and visible progress is what gives users a reason to come back.

Sharing an imperfect MVP early taught me more than polishing privately would have. Adoption and positive feedback are meaningful signals — they are still not proof of academic improvement.

Reusable content structures are what make it cheap to go from one module to many.

Stating these plainly is part of the work. These are the boundaries of what the evidence here can and cannot support.

The gap is measurement. Adoption is known; behaviour is not.

ExamMaster taught me that the reusable product was not a specific Computer Science question bank; it was the structure for turning fragmented learning material into a guided practice experience.

That insight later influenced experiments around AI-curated Product Management learning, lecture and video resource curation, and interview preparation. Those are a future product direction and a product insight — not part of the validated ExamMaster MVP.