Doppel is a startup for detecting and taking down social engineering attacks for brands. They crossed Series B funding

The Onsite Process

• A 4 hour call session with the time breakdown as:
  a. 15 min for intro/overview of the exercise
  b. 45 min coding up boilerplate in front of interviewer
  c. 2.5 hour solo coding additional features. If you have questions for the interviewer, you can email or text and they'll answer. They just won't sit in the zoom call in this portion!
  d. 30 min code review where you walk through how you spent the 2.5 hours and answer questions about the design

• Open book/open internet, but code must be written in Coderpad b/c some of the data you need will be provided that way

• You are allowed to use AI. Doppel provided Claude Pro, so I took advantage so I didn't have to pay for it myself.

What the heck did you need to build

The feature I needed to build was a phishing detection engine! AKA, "mini-Doppel"

There is a google doc given to you about phishing detection engine and a walkthrough of what data you are given.

You need to go through some labeled data and HTML files and identify if it's a "look-alike" to a "real brand"

Advice

• Use Claude Pro from Doppel. I was able to one-shot the boilerplat for the 45 min section and then spent time with the interviewer to align on how I want to spend the 2.5 hour solo coding time. Make sure you understand the codebase you're given and ask all the questions you need

• Read through the requirements doc carefully. You want to make sure that you're designing a simple solution that works and can be extendd upon. Don't fall into trap of taking complicated route especially if you have to justify your design choices to the interviewer

• This interview is more about practical engineering skill rather than leetcode brainteaser. Try not to overthink