As computer science students applying for internships, we noticed that interview preparation is hard to do well when it is not clear what companies will be asking for next. We wanted to make practice easier to get to and more forward-looking than drilling the same list of questions.
We built Talk To Future, an interview simulator that pulls from current industry trends and academic research to figure out which technical skills are becoming relevant. It generates interview questions from that, asks follow-up questions based on how you answer, and gives feedback you can actually act on.
Our biggest challenges were getting real-time voice conversations working, building a question generation system driven by the research we pulled in, and making the AI evaluations consistent enough to trust. Along the way we picked up hands-on experience with AI APIs, backend architecture, and full-stack development.