Case study / 02
AI CAS Mock Interview Platform
The coaching had to survive without the coach.
- Role
- Led the build
- Surface
- Student-facing practice platform
- Worked across
- Product · Frontend · AI
An AI coach for the UK CAS interview. Answers are transcribed and scored against criteria built with visa counsellors, then returned as personalised feedback — content, clarity and fluency, answer by answer.
01From practice to actionable feedback.
Students could take a CAS mock interview whenever they needed to practice. Their recorded answers were transcribed and evaluated against criteria developed with visa counsellors, turning each session into immediate, structured feedback.
02Practising for a CAS interview meant waiting for someone else.
A student who wanted to rehearse had to find a counsellor with time, sit the interview live, and then wait again for that counsellor to write up what went wrong. Every round of practice cost two people a slot in their calendar — so most students got very few rounds.
Before
- Schedule a counsellor
- Wait for the slot
- Attend the interview live
- Wait two to three days for feedback
- Schedule again to practise
With the platform
- Take a mock interview any time
- Answer realistic CAS questions
- Transcript generated
- Evaluated against counsellor-informed criteria
- Immediate overall and question-level feedback
- Practise again
The bottleneck was never the questions.
It was everyone’s calendar.
03I led the platform with the people who run the real interview.
I led development of the mock interview platform alongside visa counsellors, users, managers and backend engineers — defining the interview flows, the evaluation rubrics, the transcript-based feedback and the product improvements that followed from watching it get used.
The rubric had to come from the people who run the real thing.