AMD interview questions

Showing 1–19 of 19 results

• Build a UVM testbench: driver, monitor, scoreboard and coverage-driven closureEasy

  A hardware design verification question on building a SystemVerilog UVM testbench for an RTL block. It covers agents, drivers, monitors and scoreboards, checking results that may leave out of order, catching missing outputs, and using code and functional coverage to steer constrained-random stimulus toward sign-off.

  AMDSoftware EngineerSoftware Engineering FundamentalsTechnical Screen
• C++ Code Reading: Bug in a Linked-List insertAfter and Predicting Polymorphic OutputMedium

  Two C++ code-reading exercises: find the bug in a singly linked list insertAfter function that is missing a pointer update and predict what the program prints, then predict and explain the output of code mixing virtual overriding, name hiding, constructors, default arguments and pass-by-value. It tests precise tracing of pointers and dispatch rules.

  AMDSoftware EngineerSoftware Engineering FundamentalsOnsite
• Compare Single-Agent and Multi-Agent DesignsMedium

  Compare single-agent and multi-agent systems through task independence, specialization, review isolation, coordination cost, shared-state control, and evidence-based integration.

  AMDSoftware EngineerML System DesignOnsite
• Design and Implement a Fixed-Capacity LRU Cache with Constant-Time OperationsMedium

  Design and implement a fixed-capacity least-recently-used cache whose get and put operations run in constant time, while answering follow-up probes during coding. It tests data structure composition, careful iterator handling, eviction edge cases, and extensions such as thread safety and expiry.

  AMDSoftware EngineerSoftware Engineering FundamentalsOnsite
• DevOps Experience, Jenkins Pipelines and Plugins, and Reading Very Large Log FilesMedium

  Practical DevOps questions: describe your DevOps experience, explain how Jenkins pipelines, agents and plugins are used to build and test software, and show how to fetch and search a log file too large to open or load into memory. It tests hands-on CI/CD knowledge and efficient handling of large files.

  AMDSoftware EngineerSoftware Engineering FundamentalsOnsite
• Evaluate Agents That Generate or Optimize Kernels and Compiler CodeMedium

  Evaluate kernel and compiler agents with bounded transformation spaces, isolated execution, differential correctness checks, numerical tolerances, representative benchmarks, and reviewable evidence.

  AMDSoftware EngineerML System DesignOnsite
• Explain Cache Conflicts and Related Memory-Access ProblemsMedium

  Explain cache-set conflicts with address mapping, then distinguish eviction from false sharing and GPU bank conflicts so padding and layout changes target the right cause.

  AMDSoftware EngineerSoftware Engineering FundamentalsOnsite
• Explain GPU Register Spills and Their Performance CostMedium

  Explain GPU register spills through live-variable pressure, compiler allocation, spill traffic, occupancy tradeoffs, and measured optimization choices.

  AMDSoftware EngineerSoftware Engineering FundamentalsOnsite
• Explain How You Learn New Technical TopicsMedium

  Explain a practical technical-learning process through a defined knowledge gap, authoritative sources, small experiments, feedback, and verification before real use.

  AMDSoftware EngineerBehavioral & LeadershipOnsite
• Explain Skills and Progressive Disclosure in an AgentMedium

  Explain reusable agent skills, prompt context, progressive disclosure, selection metadata, referenced resources, and the boundaries of permissions and validation.

  AMDSoftware EngineerSoftware Engineering FundamentalsOnsite
• Explain the ReAct Agent LoopMedium

  Explain a ReAct tool loop through state, action selection, observations, recovery, stopping criteria, and validation of task completion.

  AMDSoftware EngineerMachine LearningOnsite
• Fast Retrieval at Scale: Databases, Indexing vs Sharding, Red-Black Trees and TriesMedium

  Conceptual questions on data storage and retrieval: what a database is and how to find data quickly in a very large dataset, how sharding differs from indexing, and how red-black trees and tries work. It tests understanding of access paths, horizontal scaling and core search data structures.

  AMDSoftware EngineerSoftware Engineering FundamentalsOnsite
• Find the Largest K Elements with PartitioningMedium

  Find the largest k occurrences with partition-based selection, sort only the output, and compare quickselect pivot risks with heap and full-sort alternatives.

  AMDSoftware EngineerCoding & AlgorithmsOnsite
• How and How Often You Use AI Tools in Day-to-Day Engineering WorkMedium

  A behavioral question asking where you use AI tools in your engineering work, how often, and how much you rely on them. It tests judgment about when AI helps or hurts, how you verify generated output, and awareness of confidentiality and code quality risks.

  AMDSoftware EngineerBehavioral & LeadershipOnsite
• Optimize the Memory Accesses of a GPU Matrix TransposeMedium

  Analyze row-major GPU transpose accesses and use tiled shared staging for coalesced reads and writes, with barrier safety, edge handling, and device-specific bank checks.

  AMDSoftware EngineerSoftware Engineering FundamentalsOnsite
• PCIe deep dive: TS1/TS2, LTSSM link training, TLP types and transaction ordering rulesEasy

  A PCI Express protocol deep dive for a verification role. It covers TS1 and TS2 training ordered sets, LTSSM link training from Detect to L0 and through Recovery, TLP headers and the Posted, Non-Posted and Completion categories, and the ordering rules that decide when one transaction may pass another.

  AMDSoftware EngineerSoftware Engineering FundamentalsTechnical Screen
• Profile a GPU Workload and Choose a Parallelism StrategyMedium

  Profile a GPU workload from its critical path to kernel counters, then choose parallelism based on independence, memory capacity, communication, and measured end-to-end gains.

  AMDSoftware EngineerSoftware Engineering FundamentalsOnsite
• Trace an LLM Request Through a Paged-KV Inference EngineMedium

  Trace LLM inference from request admission through prefill and decoding, explaining paged KV block tables, continuous batching, cache lifetime, and latency tradeoffs.

  AMDSoftware EngineerML System DesignOnsite
• Use the Roofline Model to Analyze Prefill and DecodingMedium

  Apply roofline analysis to LLM prefill and decoding using arithmetic intensity, sustainable bandwidth, batching, KV-cache traffic, and measured kernel behavior.

  AMDSoftware EngineerSoftware Engineering FundamentalsOnsite