Stack: Next.js, NestJS, PostgreSQL, Anam, swappable LLM scoring layer.
Try it: https://t.co/pQ9QJf2bJ3
Code: https://t.co/j0VyoiGBFi
Sessions capped at 3 min so shared Anam credits go around. Clone it, add your own keys, set any length you want.
I used to go blank in interviews.
Not from lack of experience. Someone would ask me to describe a time I handled something hard, and I'd start rewriting the story mid-sentence.
So I built an AI interviewer to practice against. Live video, real voice. ๐งต
The interviewer is a photorealistic avatar from @anam_ai. You talk to it over real video and voice, like a call.
I added one thing the spec didn't ask for: when the session ends, every answer gets scored against STAR by an LLM.
You see which parts were weak, not just "good job."
Then I saw challenge #128 on @johncrickett's Coding Challenges and it clicked immediately.
It also turned out to be a real backend workout. Real-time session handling, streaming, and a scoring pipeline pushed my NestJS further than another CRUD app would have.
What fixed it for me was the STAR framework. Situation, Task, Action, Result.
Once I had a shape to hang an answer on, I stopped improvising under pressure.
But knowing the framework and using it while nervous are two different skills. Only the second one needs reps.
We have 3 major religion in Nigeria na one religion we always criticise our heads to call out the government, wetin do the remaining religion dey no get heads ?