Microsoft isn’t just looking at how you use AI anymore; it's also checking whether you know if what it gave you is any good. This evaluation shows up on both the SWE and PM sides, but not in the same way, so let’s break each down.

SWE Interviews: Someone in our community who recently got the Microsoft AI SWE intern offer said they applied as soon as the role was posted. They described three rounds: two LeetCode-style rounds at easy-to-medium difficulty and one AI-assisted round. In the AI round you can use any LLM, and you share your screen. 

According to them, most of the interview was around requirements & edge cases and debugging whatever the AI produced. We’ve gotten a ton of questions around this round, and it looks like it’s designed to make sure you can’t coast.

PM Interviews Look Different: Another person in our community who went through the PM intern process described a recruiter screening first with 3-4 behavioral questions plus explaining why you’re interested in Microsoft and a PM role. Then came a “superday” of three 45-minute interviews: a deep dive into last year's internship, a product case, and a straight behavioral round. They didn’t have a separate AI round, but all three interviewers asked about their use of AI (and the last one spent about 10 minutes on it).

A Microsoft PM interview prep email we saw tells a slightly different story. It lists three interviews: AI Fluency (Vibe Coding), Product Sense, and Execution & Behavioral. The AI Fluency round has you use your preferred AI tools to turn an open-ended product problem into a working prototype. The email says they're assessing how you “evaluate and improve the output while retaining judgment and ownership of the result.” The other two rounds don’t allow you to use AI at all, and results come within two weeks of your interview day.

The screenshot didn’t indicate which PM role or level that email was for, so we can’t tell if the gap with is because it’s a different team, level, or format. Treat both as real until we hear more. 

All in all, the common thread is that where AI is allowed… the thing being judged is you, not the tool. Where AI isn't allowed, it really isn't, so you need to be solid with and without it. Remember that this comes from a small number of reports, so it doesn't mean every Microsoft team runs it this way.

For SWE:

For PM:

As we said in the last newsletter, don't wait around for a referral. Meta undergrad referrals don't seem to exist anymore. And for most undergrad roles, a referral doesn't carry much weight anyway, so waiting on one usually costs you more than it gets you. Get your application in first.

If the site throws a "Cannot submit" error when you try, try again later in the day. It happens every year when the roles go live, and coming back doesn't hurt your chances. One suggestion going around is to create a Meta account, or connect your Facebook account in the profile section, before you apply. We've heard of this working for others too.

The application asks the same two questions as last year. The first is what you hope to gain from the Ignite Internship if you're selected, answered in a few sentences. The second is how AI affects your life today and how you see its potential to create positive change, answered in one or two paragraphs.

Someone who got an intern offer said they had 25 minutes to implement a bug fix and 25 minutes to add features. The interviewer told them he didn't care if they finished. He wanted to see what decisions they'd make taking the code to production, so it was a mostly high-level conversation. Their tip was to show good agentic engineering practices, like not one-shotting the feature and using plan mode. Same idea as Microsoft: the process mattered more than the finished feature.

Someone who got a recruiter outreach said their process was one DSA round followed by Palantir's Learning or Decomp interviews, with no AI portion. Learning gives you a new abstract concept, like an API call, and you apply it to a problem. Decomp gives you a very basic one-liner problem, and you design the solution start to end, with what data you'd need and the technical design. It's usually no coding and pretty open-ended. Someone else reported a hiring manager resume roast session.

That's one person's report, not something Pinterest has published, but it's a good reason to get your application in. If you haven't applied to the Pinterest roles yet, we’ve posted them below for you.

Cohort: Internship

Pinterest is seeking a Software Engineer Intern for 2027 cohorts in their San Francisco, CA office paying 8,250-11,000 monthly. They are looking for people with non-academic coding experience like hackathons, code challenges, personal projects, GitHub, open source, volunteer coding experience, conference participation, etc.

Apply to Pinterest →

Cohort: New Grad

Pinterest is seeking a new grad software engineer graduating by June 2027. Different positions are fully remote or hybrid based out of their San Francisco, CA office. The salary range is 129K-174K, and all positions are US-only.

Apply to Pinterest →

IBM is hosting a virtual interview prep event today at 5:00pm EST. The session covers live interview demonstrations and feedback, plus how successful candidates should structure their answers and how interviewers evaluate them.

Explore the IBM event →

Hack Western 13 is running an in-person university hackathon on November 20-22, 2026. Team applications close on October 18. Full details are on the website.

See Hack Western →

If you've applied to or interviewed with Pinterest this cycle, reply with the role, where you are in the process, and what each step looked like. If you've gotten an offer or a return offer, tell us the role and, if you're comfortable sharing, the comp.

Subscribe to Direct Consideration to get the urgent roles, recruiter intel, events, and opportunities worth knowing about as they happen.