I solved the task with Claude 3.7 and got rejected by an AI startup

[hedin_hiervard]

I was interviewing for a senior software engineering role at a startup. They assigned a take-home algorithmic problem with a presentation, estimating about half a day of work.

For clarity: no restrictions were placed on tools, only on the deliverables. In previous technical interviews with other companies, I'd openly used AI—even running ChatGPT during live calls with engineers. Everyone seemed genuinely intrigued by this approach.

Could I have solved it manually? Certainly. Research the problem, internalize the domain knowledge, prepare the presentation... a few hours of work.

Or I could collaborate with Claude 3.7 and finish in ~30 minutes through quick iterations, including the presentation, a quick summary of the solution, and speaker notes. I chose the latter path.

They rejected me because of this.

The irony? This was an AI startup.

[ottaborra]

Something I've learnt the hard way. While there is merit to be able to do things without model assistance, this is a tale as old as time: purists who resist model usage will in the end be the unintentional enforcers of a culture of: everybody uses AI but the ones who come on top are people who pretend otherwise/who can best hide the fact that they are using them

You're probably aware of things like hyping up your resume, hyping up your stories and the existence of people on the other side who know of this dog and pony show and continue to play along? What you're going through in my opinion is the AI age analogue of that: everyone probably uses them but the people who come on top are the people who are able to pretend they don't

[theamk]

In real job, especially in the senior software engineering role, you are expected to know _a lot_ of knowledge specific to the company - "ETL job works fastest with narrow tables", "our backend struggles with Foo table, so it's better to minimize the access...", "latest version of componentX is buggy", and so on. The kind of things AIs, which only have the stale public internet knowledge, cannot handle. That's why they are hiring people after all.

Of course it's impossible to test the company-internal knowledge during the interview, so there are approximations. The company that wanted to hire you gave you a knowledge test using public data. You did not demonstrate that you can internalize knowledge, so you failed.

(FWIW I think that take-home tests were rather useless, because of the cheating, and they became even more useless in ChatGPT era. Today, if a company wants to test candidate's ability to learn and apply new things, they'll need to come up with a new test. Sadly, I don't know a good solution for this)

[minimaxir]

This is a case where before starting it's more pragmatic to be explicit and proactively ask "I will be using Claude for these assignments because it gives me a productivity increase. Is that ok?" (although there is a risk that merely asking that question will cause them to think you are using AI regardless)

There's still stigma around using AI to make jobs easier, even if it's a net productivity increase. In interviews, it makes the hiring managers internally ask the question "why do we need to pay 6 figures for this senior engineer when they are just prompting Claude like my parents can?"

[rvz]

> They rejected me because of this.

Unless they explicitly told you that you cannot use AI in the rejection letter, they won't tell you why you got rejected. Or maybe they chose another person that used AI to get the position.

We don't know and every company has different requirements and there is no point risking it with AI.

> The irony? This was an AI startup.

Are they really an "AI startup" or just a wrapper onto someone-else's AI model? (AI powered)

Given the position and you being good with AI, I think you should clone their entire startup and compete against them.

[hakaneskici]

Did they elaborate why? What do you think they were expecting from you that you didn't demonstrate?

Best wishes with your job search.