← Back to Directory

NVIDIA interview notes, page 12 of 17

FAANG / Big Tech

Most of these NVIDIA write-ups come from Software Engineers (137), then Product Managers (15) and Data Scientists (10). The most reported rounds are the technical phone screen (106), the system design round (28), the hiring manager screen (13) and the product sense round (13).

166 interview notes · updated Aug 2026

NVIDIA·Software Engineer·Technical Phone Screen

Jun 2026
SQL-heavy technical screen for a software engineering role at NVIDIA. Just one question but they really dug into it, including a follow-up about ambiguous data that I wasn't fully prepared for.
  • You have four relational tables: country, state, city, and zip (which stores population per ZIP code). Given a city name as input, write a SQL query that returns the total population for that city. Then explain how you'd handle the case where multiple cities share the same name across different states or countries.

“The join chain itself wasn't bad.”

View Post

NVIDIA·Software Engineer·Technical Phone Screen

Jun 2026
NVIDIA technical screen for a software engineer role, focused on quantization math. Two follow-up questions on asymmetric int8 quantization after an earlier symmetric version. Pretty niche stuff, not your typical coding round.
  • If you introduce a zero point into FP32 to Int8 quantization, how do you compute it and apply it? Implement the full conversion using both a scale and a zero point.
  • When the input distribution is asymmetric (min and max not centered around zero), how do you pick scale and zero point to fully use the int8 range? Fold both into quantize and dequantize functions and explain why the result is correct.

“I knew the symmetric case cold, but adding the zero point tripped me up at first.”

View Post

NVIDIA·Software Engineer·Technical Phone Screen

Jun 2026
NVIDIA software engineer interview that went deep into matrix multiplication, from complexity analysis all the way down to CUDA tiled GEMM. Heavy on systems-level thinking and GPU architecture awareness. Not a typical coding round.
  • What are the time and space complexity of the naive triple-loop matrix multiplication algorithm for two N x N matrices?
  • Can you describe Strassen's algorithm and other sub-cubic matrix multiplication approaches at a high level?
  • What is the memory/cache bottleneck in naive matrix multiplication, and how does blocked or tiled matmul address it?
  • Walk through a tiled GEMM implementation on GPU with CUDA, including how you load tiles into shared memory, synchronize threads, and accumulate the output tile. How do you choose tile size?

“Easy warmup but i second-guessed myself on the space complexity for a second.”

View Post

NVIDIA·Software Engineer·Technical Phone Screen

Jun 2026
Interviewed for a SWE role at Nvidia, got asked a pretty foundational systems question about memory leaks. Short and to the point, not much else to report.
  • What is a memory leak and what are some ways to prevent it?

“Knew this one but still fumbled the prevention side a bit.”

View Post

NVIDIA·Software Engineer·Technical Phone Screen

Jun 2026
NVIDIA technical screen for a software engineer role, focused on low-level numerical computing. One question, but it had a lot of surface area and they clearly wanted you to go beyond just writing the code.
  • Given an FP32 tensor and a floating-point scale factor, implement a function that converts each element to Int8 by dividing by the scale and rounding. Then walk through the numerical issues: rounding mode choices, overflow behavior, and how you'd handle saturation to the [-128, 127] range.

“The coding part wasn't the hard part.”

View Post

NVIDIA·Data Scientist·Technical Phone Screen

Jun 2026
Deep technical screen for a Data Scientist role at NVIDIA, all centered on image classification in a healthcare context. Five heavy questions back to back covering everything from overfitting diagnosis to cross-validation design. Felt more like a written exam than a conversation.
  • How do you rigorously define overfitting in an image classification context, diagnose it using learning curves and calibration, and what remedies would you propose while justifying the trade-offs?
  • Explain DenseNet's connectivity pattern, growth rate, bottleneck and transition layers, and how its parameter and memory complexity compares to ResNet. When would you prefer it, and can you estimate the parameter count for a small configuration you define?
  • Design a full data preprocessing and augmentation pipeline for medical images, address class imbalance, and walk through the common data leakage traps and how you'd catch them.
  • What is the difference between model hyperparameters and learned parameters? Describe a hyperparameter tuning strategy including search spaces, budget constraints, early stopping, and how you'd make the whole process reproducible.
  • Design a patient-level cross-validation scheme that prevents leakage across patients, scanners, and time. How do you aggregate metrics with confidence intervals and compare two models fairly?

“I started with the bias-variance framing and that felt solid, but when they pushed on calibration specifically I kind of stumbled.”

View Post

NVIDIA·Software Engineer·Onsite - System Design / Architecture

Jun 2026
System design round at NVIDIA for a software engineering role. The whole session was basically one big question about URL shortening, but it sprawled into a lot of sub-areas fast and I felt like I was playing catch-up by the midpoint.
  • Design a URL-shortening service like TinyURL or bit.ly. Walk through your functional requirements, how you'd generate short keys, your storage choices, caching strategy, and how you'd handle abuse.

“I started with requirements which felt fine, listed the obvious stuff like shorten, redirect, custom aliases, expiration.”

View Post

NVIDIA·Data Scientist·Onsite - System Design / Architecture

Jun 2026
System design round at NVIDIA for a Data Scientist role, focused entirely on designing a production-grade HTTP API for serving image model predictions. The question was sprawling and covered basically everything: auth, observability, versioning, security, and even an OpenAPI snippet. A lot to hold in your head at once.
  • Design an HTTP API to serve image-based model predictions, covering endpoint design, request/response schemas, auth, rate limiting, security, observability, backward compatibility, and provide an OpenAPI 3.0 snippet for one endpoint.

“This question is basically five questions stapled together.”

View Post

NVIDIA·Software Engineer·Technical Phone Screen

Jun 2026
Interviewed for a SWE role at Nvidia and got asked about callback functions. Short and to the point, nothing too wild.
  • Can you explain what callback functions are and how they work?

“Pretty standard JS concept but I fumbled the explanation a bit.”

View Post

NVIDIA·Software Engineer·Technical Phone Screen

Jun 2026
NVIDIA software engineer interview with a heavy focus on sorting algorithms, complexity analysis, and knowing when the textbook rules break down. More theoretical than I expected for an SWE role.
  • Walk through the common sorting algorithms, their time and space complexities across best, average, and worst cases, and identify which are stable, in-place, and comparison-based.
  • Why is quicksort O(n²) in the worst case, and what triggers that scenario?
  • Why is merge sort guaranteed O(n log n) in the worst case when quicksort is not?
  • Under what conditions can non-comparison sorts like counting sort or radix sort beat the O(n log n) lower bound, and why doesn't that lower bound apply to them?

“This felt like a trivia dump at first but it got tricky fast.”

View Post