Katkat

Company

NVIDIA interview questions

95 practice questions in the style of NVIDIA's interviews, covering Data Manipulation (SQL/Python), System Design, Software Engineering Fundamentals, Machine Learning, Coding & Algorithms, Analytics & Experimentation, Behavioral & Leadership, ML System Design, Statistics & Math.

Questions95

6 easy67 medium22 hard

Your progress0/95

NVIDIA interview questions by the numbers

Last updated

Katkat has 95 NVIDIA interview questions. The largest topics are Coding & Algorithms (25%), Software Engineering Fundamentals (24%) and System Design (19%). By difficulty: 6% easy, 71% medium and 23% hard. The most common round is Technical Screen (54%). Frequent concepts include Distributed Systems, System Design, Data Structures and Algorithms.

Topics

Difficulty

  • Easy6% · 6
  • Medium71% · 67
  • Hard23% · 22

Interview rounds

  • Technical Screen54% · 51
  • Onsite23% · 22
  • HR Screen11% · 10
  • Online Assessment9% · 9
  • Other3% · 3

Roles

  • Software Engineer82% · 78
  • Data Scientist11% · 10
  • Data Engineer3% · 3
  • Applied Scientist2% · 2
  • Machine Learning Engineer2% · 2

Frequent concepts

  • Distributed Systems5% · 5
  • System Design5% · 5
  • Data Structures4% · 4
  • Algorithms3% · 3
  • Greedy3% · 3
  • Optimization3% · 3

Most mentioned in interview experiences

  1. 1.Discuss activities, role, and project deep dive5 reports
  2. 2.Compare deep learning framework trends2 reports
  3. 3.Clarify and Design a Streaming Top-K Tracker1 report
  4. 4.Compare Java and Python: Execution, Typing, Concurrency and When to Use Each1 report
  5. 5.Count Distinct Strings With a Hand-Built Hash Table, Then Parallelize It for GPUs1 report
  6. 6.Define developer-centric usability metrics1 report
  7. 7.Demonstrate cultural fit and sales-oriented leadership1 report
  8. 8.Describe model-to-GPU execution pipeline1 report
  9. 9.Design a Key-Value Store, From the Single-Node Storage Engine to Replication1 report
  10. 10.Design a Task Scheduler That Runs Each Task Reliably When It Is Due1 report

Counted from 16 candidate interview experiences on Katkat. How we collect and count

Showing 91–95 of 95 results

  • Solve unique triplets summing to targetMedium

    Solve unique triplets summing to target evaluates algorithm design, data structures, correctness, complexity, edge cases, and implementation details in a realistic interview setting. A strong answer states assumptions, handles edge cases, explains trade-offs, and shows how to validate the result clearly.

    NVIDIASoftware EngineerCoding & AlgorithmsTechnical Screen
  • Streaming Top-K Frequent Elements with O(1) Updates Instead of a HeapMedium

    Design a structure that consumes an unbounded stream and reports the K most frequent elements at any time, first with a heap and then with constant-time updates after the interviewer rejects the heap. It tests per-operation cost analysis, data-structure invariants, deterministic tie handling, and extensions to sliding windows, approximate counting, and sharded streams.

    NVIDIASoftware EngineerSoftware Engineering FundamentalsOnsite
  • Verification Plan for a FIFO, a Round-Robin Arbiter and a Valid/Ready HandshakeHard

    A hardware design-verification interview question: plan how to verify a FIFO, a round-robin arbiter and a valid/ready handshake interface. It tests testbench architecture with scoreboards and reference models, constrained-random stimulus, corner cases, SystemVerilog assertions and functional coverage.

    NVIDIASoftware EngineerSoftware Engineering FundamentalsOnsite
  • Write SQL to sum city population by nameEasy

    Assesses SQL proficiency in relational joins and aggregation by asking for a total population computed from ZIP-level data and evaluates reasoning about duplicate entity names.

    NVIDIASoftware EngineerSoftware Engineering FundamentalsTechnical Screen
  • Write a Tiny Linux Resource Reporter That Flags Usage SpikesEasy

    Write a small Linux agent that reads live CPU, memory and disk usage from kernel-exposed files, reports them as JSON to a collector endpoint over HTTP, and adds a rolling-window check that only flags a sustained resource spike rather than a single noisy reading.

    NVIDIASoftware EngineerSoftware Engineering FundamentalsOnsite

NVIDIA interview experiences

All 23 →
  • GPU Hash Partition Phone Screen in C++, Two Weeks to the Next RoundNVIDIA · Software Engineer · Senior+ · Technical Screen · Oct 2026
    In progress

    I asked a friend to refer me to a lot of NVIDIA positions. This was the first one to respond, and I really felt how slow the process is. After the phone screen it took two weeks just to schedule the next interview. The interviewer was a bit of a nerd. The topic was GPU partitioning, in C++. Given an array and P partitions, do a hash partition, and the output has to return the offsets. It is the c…

  • Three Onsite Rounds, Rejected Three Days LaterNVIDIA · Software Engineer · Onsite · Oct 2026
    Rejected

    After I passed the one phone screen round, they told me to go straight to the onsite. There were only three rounds in total, which made me realize I wasn't going to be that lucky, and sure enough, three days after the interview I was told I was rejected. The first round was straight to a director: BQ plus a deep dive. He felt like a smiling tiger, rambling about this and that, and then suddenly s…

  • Three System Design Rounds and a Director RoundNVIDIA · Software Engineer · Onsite · Oct 2026
    In progress

    I interviewed last Friday. There were four rounds, and in the first three the interviewers were all senior SDEs, with one system design question per round: A task scheduler Implement a version-controlled download service. For example, if you want to download python2.7 and you find that your machine is missing a few of its dependency packages, you have to download those as well. I thought about th…

  • Verification Role, Offer After Eight WeeksNVIDIA · Software Engineer · New Grad · Onsite · Sep 2026
    Offer

    The interview mostly revolved around FIFO, round-robin arbiter, valid/ready handshake, scoreboard/reference model, constraints, SVA and functional coverage. They give you a small design/spec and have you discuss it from the angle of verification architecture, corner cases, assertions and coverage. They also ask brain teasers, and there are a lot of them online. The coding/RTL part is mainly basic…

  • GPU Search Engine Team Phone Screen on Counting Distinct StringsNVIDIA · Software Engineer · Technical Screen · Jul 2026

    This is for a team at Nvidia that is turning an open-source search engine into a GPU implementation. I personally have no hardware or GPU experience at all. They replied the day after I applied and said they wanted a phone screen, and I found out early on, after looking into the background of the interviewer, that the interviewer was a hardware engineer. I assumed they cared about my search engin…

More companies