Updated by Cursor candidates
Software Engineer (New Grad) Interview Experience
They basically gave me access to part of their codebase and were like, "Figure it out, see anything you want to build, and just build it." For eight hours I worked out of a Slack group, then presented the feature at the end of the day.
Interview process
The whole process felt way more startup-y than normal. First was a 45 minute informal chat where he went through my background, asked why Cursor, and was very candid about the company and pace. Then I had a 60 minute repo-based coding round where I implemented a hash tree they actually use, and I was allowed to use Google, GPT, and Cursor for targeted syntax help. The final was the most unusual part: an 8 to 9 hour remote onsite where they gave me part of the codebase, a Slack channel, and asked me to figure it out, build a feature autonomously, and present it. They told me they're still figuring out what new grad hiring even looks like, so the process felt unconventional and a little vague, but also much more like real work than the standard interview loop.
- Recruiter screen
- Technical interview
- Final round
Interview tips
If I were prepping again, I would not just grind LeetCode and call it a day. That helped for the initial technical round, but the final was much more about whether I could jump into an unfamiliar repo, scope something reasonable, use AI tools well, and explain my trade-offs. I would practice by finding open source repos that match their stack or the kinds of features Cursor builds, give myself 30 to 60 minutes to understand the codebase, then 2 to 3 hours to scope and build a small feature. I would also be careful about asking for help the right way: use docs, README files, and AI tools first, then ask only when it is really specific to their setup. Biggest thing is just be ready to be surprised and stay on your feet.
Company culture
My read is they are hiring like a fast-moving startup, not a mature company. For this process, there was basically no recruiter layer, just a hiring manager who was very open about the pace, long hours, and the fact that people may work weekends. The org felt flat, energetic, and spontaneous, with people trying to ship updates every day or two, and the interviews matched that because they focused on autonomy, initiative, and whether they would actually want to work with me. It also seemed like they are mostly staffed with senior engineers right now and only recently experimenting with mid-level and entry-level hiring. The other big thing is they are one of the only companies I have seen actually let candidates use AI tools the way engineers would use them on the job.
Questions asked
Overview
The final was an 8 to 9 hour remote onsite with a kickoff, a Slack channel, a mid-day check-in, and then a 30 minute presentation at the end. They gave me access to a small slice of the codebase, not everything, and wanted to see how autonomously I could work. It was the most unusual interview I have done and felt the closest to actually working there.
Question types asked
Specific questions asked
Here is part of our codebase. Figure it out, choose something you want to build, and present it at the end of the day.
How did you scope the feature?
Why did you choose that feature?
Why did you add onto this existing service?
Why would this make a difference?
I explored a few backend services, mostly around how they connect to different APIs and models, and picked an add-on to one existing service. My goal was to show I could find a place to create value, scope it tightly, and ship something reasonable in a day. I used the docs, READMEs, and AI tools first, then only asked in Slack when it was something niche to their codebase. In the presentation I explained the scope, why I chose it, and why I thought it would actually improve the product.
Why did you implement it this way?
Why use this service?
Is that already in our codebase?
If this is new, what are the benefits and trade-offs?
This was the harder part. They pushed on every engineering choice, especially when I used something new instead of only following what already existed in the codebase. I had to justify why a certain service or method made sense, what the upside was, and what trade-offs I was making. My read was they know AI can help people code fast, so they were really testing whether I understood why I made each decision.
Get full access with a membership, or share your experience to try it free.
