NVIDIA Data Scientist Interview Experience — Two Screens, a Six-Round Virtual Onsite, and an Unofficial Offer the Same Day

My girlfriend has a friend who works at NVIDIA, and they referred me. About two weeks later I got my first round video interview.

I want to talk about the mindset and the prep process for the whole interview loop — if that doesn't interest you, feel free to skip straight to the interview notes below.

The most important thing: humble yourself and ask others for help (don't be shy about it — if you want to learn from people you have to spend time building the relationship; this isn't about getting someone to coach you on the surface, it's about genuinely asking for guidance). During my prep, before every single round I talked to two or three people — I mean every round — and just talking with them about the company, the role, and the interview took at least 7-8 hours total (and that's just to get started). All of them either knew the role well or had worked at NVIDIA themselves.

I never met the friend who referred me in person — we only talked on the phone once, for about an hour. He was great, and walked me through the role in much more detail than the job description did. (If you spend the time to really understand the company/role, you can tell the interviewer you went and asked people about it, which shows you're serious and committed.)

Every time I got an interview invite, they'd tell me the interviewer's name, and I'd always look them up on LinkedIn — hobbies (cooking, donating blood, sports, gaming, neuroscience, signal processing), work history (whether they'd changed fields), where they went to school, whether they were born in the US, whether their undergrad and master's were in the same field, activities outside work, etc. — to get a read on the interviewer's personality and how they talk.

The reasoning: 1. You end up knowing more about the interviewer than they know about you, which gives you confidence. 2. Once you've profiled the interviewer, it's easier to connect with them, bring up topics they're interested in, ask questions they'd find interesting, and keep the conversation flowing — they'll walk away with a good impression of you (like you two just clicked), and they'll know you came prepared.

For every piece of work experience, every research project, every project on my resume, I prepared two versions of the story. The short version (STAR format, just walk through what you did, summarize, don't get into unnecessary detail) and the long version (if the short version lands well, the interviewer will want to go deeper and ask about the details — that's what the long version is for). For every story, both versions, I had someone with strong English go through it with me to check whether people could actually follow what I was saying.

Behavioral questions: I prepared for every possible one (elevator pitch, conflicts, customer-facing situations, most challenging project, a project you failed on, leadership, where do you see yourself in 5 years, etc.).

Since everything was video interviews, you can absolutely write out all your stories and BQ answers ahead of time and basically read off them during the interview. I almost never ran into a situation I wasn't prepared for — I wrote up 13 pages of Google Docs. Besides the stories and BQ answers, I also prepped possible technical questions, background info on each interviewer, tailored questions for each of them, and the advice everyone who helped me gave me — all so I'd have it memorized.

Timeline:

• 9/14: got the invite for the first round video interview
• 9/23: first round video interview
• 9/28: got the invite for the second round video interview
• 10/2: second round video interview
• 10/6: got the invite for the virtual onsite (6 rounds total)
• 10/15: virtual onsite (4 rounds)
• 10/16: virtual onsite (2 rounds)
• 10/16: unofficial offer over email

9/23: First round video interview
Hiring manager.
Questions: 1. Why NVIDIA? 2. Why this role?
We talked through my resume and culture fit (they care a lot about culture fit — really listen to what the interviewer says about it and give feedback in real time showing you feel the same way).

10/2: Second round video interview
Manager from the healthcare group.
We only went over one experience from my resume (he told me upfront he'd only ask about one, and the rest would be covered by other interviewers).
Questions: 1. Explain overfitting 2. Explain DenseNet 3. Explain data pre-processing 4. Explain hyperparameters 5. Explain cross-validation

10/15: Virtual onsite (4 rounds)

Round 1
A teammate on the NLP side of the team.
Went over my resume (2-3 experiences, his questions were sharp and dug in deep).
Coding: Reverse linked list
Questions: 1. Explain web APIs 2. Explain CUDA

Round 2
A manager from the sales/account side of the team (not technical — he only cared about whether you'd be a good fit for sales and could work with people).
Question: If you want to sell NVIDIA GPUs to the CEO of Walmart, how would you approach it?

Round 3
A PhD on the team who worked on CV and video analysis.
Question: How do you perform matrix multiplication on CUDA?

Round 4
A teammate from MIT, very funny — the whole interview he was laughing.
The entire round was him asking about the healthcare/medical projects on my resume that he was interested in.

10/16: Virtual onsite (2 rounds)

Round 1
A manager on the IoT/robotics side of the team, who'd spent ten years at Altera before this.
Again, entirely about the experiences on my resume he found interesting — good thing I'd prepared for all of it.
Questions: 1. Conflicts? 2. Describe a project you failed on.

Round 2
The last round was the hiring manager again.
No resume talk at all — just asking how I felt about the whole interview process and what I'd learned.
You could tell whether the hiring manager was planning to make an offer.