First coding interview: Calculate the document frequency and inverse document frequency for a body of strings. Optimization is to precompute the frequency for all words in the corpus. For the multi transform, store previously computed transforms in another map.
Frontend coding interview: Update an existing React app to generate a graph using some randomly generated values using the react-chartjs-2 library. Components used the class-based components so review component lifecycle methods, i.e. not hooks.
Backend coding interview: Use SQL to calculate predicted income for a publisher going from a free to paid model. Use that to generate a graph that you built in the frontend interview. Not really a backend interview as much as it is a SQL interview. This is an open-ended interview with no right answer.
Generic coding interview: Build a data structure that returns a range of values. For example, calling set(3, 6, ‘a’) will set indices 3, 4, and 5 to the value ‘a’. Runtime should not be linear with the number of indices being set. Calling get(0, 7) would then return [null, null, null, ‘a’, ‘a’, ‘a’, null].
System design interview: Design a scalable system that sends out emails to newsletter subscribers. Newsletters can configure when emails are sent to its subscribers with a specific date/time.
Final interviews were all in one day – about 6 hours in total. Rejected at this stage. Asked recruiter for feedback and was ghosted.
The following metrics were computed from 7 interview experiences for the Substack Software Engineer role in United States.
Substack's interview process for their Software Engineer roles in the United States is extremely selective, failing the vast majority of engineers.
Candidates reported having very negative feelings for Substack's Software Engineer interview process in United States.