Roles I interviewed for

My 2025–26 search included interviews across ML / AI, quant, software engineering, and data science:

  • LinkedInMachine Learning Engineer
  • MillenniumAI Engineer
  • Tower Research VenturesMachine Learning Engineer
  • HRTAlgorithm Developer
  • SIGQuantitative Research
  • DatabricksData Science
  • Amazon BedrockAI Software Development Engineer
  • AppleAI / ML Software Development Engineer
  • GoogleSoftware Development Engineer

These are interview experiences, not a list of offers. Below is what I learned from preparing for them; I don't share company-specific questions or processes.

I'm now looking for full-time roles and keeping those notes in 26-27 new-grad search 🚧. This page is about that first internship search.

August 2025: starting from scratch

I was starting my MSCS at Georgia Tech and had no industry internship experience. Most of college had been spent doing research. I wanted to try industry, but knew little about the roles or interviews.

I started LeetCode, behavioral preparation, and applications at the same time. At first I didn't filter much by company size or role: QR, QT, MLE, SDE, DS—I applied to all of them.

More applications seemed like more chances. Once interviews arrived, I realized that more chances didn't come with more preparation time.

Every role brought something else to catch up on

Some fundamentals overlap, but the preparation differed more than I expected. SDE coding, MLE models and projects, and quant probability and statistics weren't the same material rearranged into a different order.

I knew particularly little about QR interviews. I hadn't looked into them properly before applying, so I was rushing to prepare once I got one. That also took time away from other roles. Trying it wasn't necessarily the mistake; I hadn't allowed enough time for the attempt.

120 problems in a month, and I still didn't pass

Before one interview, I solved 120 problems in a month. I had almost no prior practice, and it felt like a substantial amount of work.

In the interview, my coding speed and approach still weren't good enough. There was an additional round, and I ultimately didn't pass. That one was disappointing.

Looking back, the count showed how much I had done, not whether I could find an approach to an unfamiliar problem and implement it reliably. I needed to revisit which methods I could actually use, rather than simply keep raising the number.

Why I narrowed it to MLE / Research Engineer

After trying several directions, I was clearer that I liked work close to both research and product: investigating a problem and seeing something built from it get used. I started focusing on MLE / Research Engineer.

That doesn't mean the other directions were worse or that exploring them was wasted effort. Trying them helped me understand what I wanted to spend time learning, and how much preparation I could realistically handle at once.

Work authorization and location mattered too

I needed to consider sponsorship. During that search, some small and mid-size company opportunities weren't a fit for that reason, and I got very few interviews in that part of the market.

That describes my experience then, not a rule about all companies of a certain size. For future applications, I still need to check the particular role and current policy. It belongs in this account because it affected my options; technical preparation wasn't the only factor.

What I'd take into the next round

Some interviewers were generous with hints and willing to discuss things properly. Others left me puzzled 😂 A few interviews still aren't enough for me to judge everyday work at a company, much less rank companies for someone else.

Starting again, I'd learn about the roles earlier, focus on work I actually want, and leave more time for steady coding practice. I wouldn't assume a popular field was something I had to try.

The next search is still happening, and these views may change: