Pondly: Practice the judgment side of data-science interviews

I built Pondly because the interviews have changed and the prep has not. Models write the code now. The interview is the part they cannot write for you: what you would check, which number is lying, and what you would do when the evidence changes. Most practice still drills syntax. A Pondly question is a conversation with a senior colleague. You answer out loud, in writing. It probes. If you get stuck, the hints climb from a few ideas to the explanation. At the end you get four ratings, each one tied to something you actually said, plus one thing to keep and one habit to sharpen. The questions are written by hand, across concepts, system design and troubleshooting, at Friendly, Standard or Tough. It is for three situations. You are going for a job you actually want. You are moving into a field where you can do the work but have not been interviewed in it. Or you just want an honest read on whether the concepts still hold. There is no free session. Each turn runs on a real model, so a pass starts at A$29 for 14 days, A$39 a month, or A$49 for 30 days with nothing to cancel. If it is not for you, ask within 7 days and that payment is refunded in full. I would like to know where a session feels off: a follow-up that did not listen, a rating that did not match what you said, or a question that is not one an interviewer would actually ask.