NVIDIA Interview Question Bank
NVIDIA interviews are highly team-specific and much less LeetCode-bank-driven than most large tech loops. System Software, Linux, Tegra, DGX AI infra, and Deep Learning roles lean toward engineering scenarios: C/C++ debugging, graph validation, GPU / inference fundamentals, small utility coding, and project deep dives; some design and product-facing AI roles add system-design or technical-marketing discussions.
73
questions
Last updated
Before you start
Showing 23 of 23
Merge Two Sorted Files with Bounded MemoryFP32 Tensor to Int8 QuantizationImplement a Reference-Counted Smart PointerTop K Frequent Elements in a Data StreamMerge Two Sorted Files with Bounded MemoryImplement a Decoder KV CacheReaching PointsCount Substrings Without Repeating CharactersMerge IntervalsCount Visible TowersSimulation-style coding problem with scaling follow-up (mini system design)Python Data Processing Task (Parse, Aggregate, and Validate)Linux Shell Scripting Task (Text/Log Processing)Compress and Decompress String (Excel-like Encoding)Find Any Duplicate in Range [1, n-1] with O(1) Extra SpaceDesign an OOP Key-Value Store with set, get, and setAllTransform an N x N Matrix by Swapping Rows and Columns (Matrix Transpose Variant)C++ Debugging and Output Prediction (Pointers, Inheritance, Multithreading)Find the N-th Prime NumberSQL Aggregation Across Country/State/City/Zip TablesEncode and Decode a List of StringsGraph API: Insert/Configure/Validate with Cycle Detection and Structural ConstraintsComputation Graph Pruning to Keep the Optimal PathTranspose a 2D Matrix (C++), Discuss Memory/Cache Trade-offsFP32 Tensor to INT8 Quantization (with Asymmetric Quantization Follow-ups)Integrate a Public GET API and Transform ResultsLongest Substring Without Repeating CharactersValidate ParenthesesTree Planting on a Grid (no adjacent trees)Aggregate Logs by Status Code (count and average latency)