You see, the interviewing stack today is fragmented. To conduct an interview, you'll have to schedule a conference call on Zoom/Google Meet/MS Teams, then have a CoderPad session, and, most recently, connect your AI note-taker to capture notes. After the interview, it is an open secret that a lot of people paste the transcripts into ChatGPT and sometimes the code too and from there try to generate post-interview notes to send to the hiring manager or debrief team.
You can already see how crazy this looks. I strongly believe that everything you need to conduct an interview should be in one place: the transcript, the code, and/or the diagram (for system design interviews). So that when you want to extract valuable insights using an LLM, you have enough context about what they said, at what time and what code they produced. This is not to say the LLM makes the decision, but it helps you analyse and get a clearer picture of how the interview went so you are making hiring decisions not based on vibes but actual data.
I built ScreenStack, an MVP as a proof of concept, and now I'm looking for design partners. If you are currently conducting coding interviews or any interview at all and would like to try it out, please shoot me an email at screenstackhq [at] gmail dot com. I'll be happy to walk you through how it works, and hopefully you use it to conduct some interviews entirely for free; my only ask is that you give me feedback.
Thank you in advance.