314,552 interview questions from 6,000+ companies.
Tests ownership during a production incident, including structured debugging, stakeholder communication, and learning from high-pressure technical problems.
Tests graph algorithm knowledge and ability to implement cycle detection correctly.
Explain how to detect vanishing or exploding gradients and stabilize deep neural network training.
Tests GPU programming skills and performance awareness for real-time perception workloads.
Tests object-detection post-processing skills relevant to perception pipelines.
Tests backpropagation fluency for convolutional networks used in vision models.
Tests understanding of detection architectures relevant to autonomous driving perception.
Tests knowledge of multi-scale feature aggregation for mapless driving perception.
Tests ability to optimize performance-critical C++ code for compute-heavy ML workloads.
Tests understanding of C++ memory behavior and its implications for latency-sensitive systems.
Tests ability to extend PyTorch with custom differentiable ops for ML training and inference.
Tests careful implementation of convolution details that affect model correctness and performance.
Tests capability to design robust high-level behavior logic for complex driving scenarios.
Tests ability to reason about performance using memory hierarchy and data layout choices.
Tests knowledge of dynamic dispatch and performance tradeoffs in C++.
Tests judgment in managing speed versus rigor for safety-critical systems.
Tests ability to optimize deep models for edge deployment while preserving safety-critical accuracy.
Tests planning algorithm selection and understanding of kinodynamic planning representations.
Tests statistical reasoning and ability to diagnose issues in linear modeling.
Tests algorithmic thinking and efficient data-structure design.
39 total questions
