AMD · OA → Phone screen → Onsite
In processAMD Interviews by Role through Oct 2026: 9 Experiences with Hiring Level Not Reported
Note: This is the most complete collection of interview information I could find in this round of searching. If friends need more information in the future, I may start another round of research. Good luck with your interviews! — 11-02-2026 (jane)
Coverage: 9 of 9 AMD Hiring Level Not Reported accounts in 9 separate role sections.
Other AMD hiring levels
- Experienced Full-Time: 16 accounts in 3 parts.
- Intern / Co-op: 18 accounts in 2 parts.
- New Grad: 11 accounts in 2 parts.
Roles in this article
- System Architect — 1 account.
- Windows Kernel Mode Driver Developer — 1 account.
- EDA Tools Developer — 1 account.
- Frontend Developer with System Design — 1 account.
- Firmware Development Engineer — 1 account.
- Software Validation and Debug Engineer — 1 account.
- Role not reported — GPU telemetry design — 1 account.
- C++ Developer — 1 account.
- Software Engineer — 1 account.
System Architect
1 account in this role.
What came up
Reported topics: coding tasks or technical assessments; projects or previous work.
01: Markham: a System Architect interview with Python coding
Interview: September 2026 · System Architect · Markham, Canada · Selected
This adjacent systems role had a roughly one-hour technical interview. The candidate described a detailed resume discussion, situational questions and a Python LeetCode problem, without naming the problem.
They reported receiving an offer in October 2026. The card's 2027 work term describes the intended placement, not the interview year.
Windows Kernel Mode Driver Developer
1 account in this role.
What came up
Reported topics: coding tasks or technical assessments; C/C++ or memory questions; OS or concurrency.
Questions we found
Algorithms / Data Structures
| Question | Frequency | All last seen dates |
|---|---|---|
| Convert a string to an integer | 1 | Reports: 2026-09 (interview) |
02: A Windows kernel-driver interview in C
Interview: September 2026 · Windows Kernel Mode Driver Developer · Markham, Canada · No offer
The 90-minute round covered OS concepts, C/C++ inheritance and casts, followed by three C tasks.
One involved bitwise operations, another asked for string-to-integer conversion, and the third involved multithreading. The exact bit transformation and thread-coordination requirements were not shared. The linked parser defines its own whitespace/sign grammar and 32-bit clamping rules.
EDA Tools Developer
1 account in this role.
What came up
Reported topics: coding tasks or technical assessments; C/C++ or memory questions; projects or previous work; database or retrieval questions.
Questions we found
Algorithms / Data Structures
| Question | Frequency | All last seen dates |
|---|---|---|
| Implement an LRU cache | 1 | Reports: 2026-08 (interview); Additional same-company practice history: 2026-10-03 (report date), 2019-10-28 (report date) |
| Repair the missing next-pointer update in insertAfter | 1 | Reports: 2026-08 (interview); Additional same-company practice history: 2026-10-03 (report date) |
| Predict method-overriding output (original code not shared) | 1 | Reports: 2026-08 (interview); Additional same-company practice history: 2026-10-03 (report date) |
Behavioral (BQ)
| Question | Frequency | All last seen dates |
|---|---|---|
| Where and how much do you use AI at work? | 1 | 2026-08 (interview) |
C/C++ / OS Questions
| Question | Frequency | All last seen dates |
|---|---|---|
| const versus static | 1 | 2026-08 (interview) |
| Fast retrieval from a large dataset | 1 | 2026-08 (interview) |
| Sharding versus indexing | 1 | 2026-08 (interview) |
| Delete through a base pointer without a virtual destructor | 1 | 2026-08 (interview) |
| Stack versus heap memory | 1 | 2026-08 (interview) |
03: EDA tools: an LRU cache and a missing linked-list pointer
Interviews: August 2026 · Report date: October 3, 2026 · EDA tools development
The three rounds were scheduled on the same day.
Round 1 — theory
Questions jumped between DevOps-team experience, const versus static, databases and fast retrieval from large datasets, red-black trees, tries, Jenkins plugins, fetching a large log, sharding versus indexing, and stack versus heap memory. One C++ question asked what happens when an object is deleted through a base pointer without a virtual destructor.
Round 2 — live coding
The candidate implemented an LRU cache, with follow-ups while writing it. The account does not preserve those follow-ups.
Round 3 — code output and debugging
An insert-after linked-list routine searched for afterVal, returned if no node matched, then allocated a new node. Its next pointer was set to the old successor, but the update from the located node to the new node had been removed: temp->next = newNode.
The linked insert-after exercise uses a deterministic list/position interface; the original bug above used value lookup.
Another task involved predicting output under method overriding and explaining the result. The original overriding code was not included, so the linked exercise is a practice completion rather than that exact snippet.
The author says there were three output-based questions, but only two distinct tasks can be recovered. The displayed insertAfter code is the same debugging task, not a fourth question. The interviewer also asked where and how much the candidate used AI at work.
Frontend Developer with System Design
1 account in this role.
What came up
Reported topics: coding tasks or technical assessments; C/C++ or memory questions; projects or previous work; OS or concurrency; JavaScript / React.
Questions we found
Algorithms / Data Structures
| Question | Frequency | All last seen dates |
|---|---|---|
| Implement a basic shared_ptr | 1 | Reports: Post displayed as 4 months old on 2026-10-30; exact date unavailable; Additional same-company practice history: 2019-10-28 (report date) |
| Implement a stack | 1 | Reports: Post displayed as 4 months old on 2026-10-30; exact date unavailable |
| Implement a multithreaded queue | 1 | Reports: Post displayed as 4 months old on 2026-10-30; exact date unavailable |
C/C++ / OS Questions
| Question | Frequency | All last seen dates |
|---|---|---|
| Polymorphism and virtual inheritance | 1 | Post displayed as 4 months old on 2026-10-30; exact date unavailable |
JavaScript / React Questions
| Question | Frequency | All last seen dates |
|---|---|---|
| Hoisting, event loop and callback queue | 1 | Post displayed as 4 months old on 2026-10-30; exact date unavailable |
| Promise internals | 1 | Post displayed as 4 months old on 2026-10-30; exact date unavailable |
| How useState, change detection and re-renders relate | 1 | Post displayed as 4 months old on 2026-10-30; exact date unavailable |
| State batching and stale state | 1 | Post displayed as 4 months old on 2026-10-30; exact date unavailable |
| What useEffect returns and when cleanup is used | 1 | Post displayed as 4 months old on 2026-10-30; exact date unavailable |
| Whether useRef triggers a state update | 1 | Post displayed as 4 months old on 2026-10-30; exact date unavailable |
04: A frontend role that also asked for shared_ptr and a threaded queue
Post displayed as 4 months old on October 30, 2026; exact date unavailable · Frontend Developer with System Design
Round 1 — JavaScript
An easy DSA task had a logical bug. The candidate fixed it after a hint, but did not share the task itself. The discussion covered hoisting, the event loop, the callback queue, promise internals and promise-output questions whose code was not preserved.
Round 2 — React
Questions covered useState, change detection, re-renders, whether state behaves like an ordinary variable, batching and stale state. They discussed what useEffect returns and when cleanup is useful, as well as Redux.
The candidate wrote a simple React component, but did not preserve its intended behavior. They also reported giving a wrong answer about useRef and whether it triggers a state update.
Round 3 — C++ and concurrency
The interviewer asked about memory management and smart pointers, then for implementations of a basic shared_ptr, a stack and a multithreaded queue. The linked shared-ownership exercise supplies an owner-operation interface: C++ uses RAII, while other languages model explicit logical closure. The queue exercise uses real synchronized workers and checks a finite batch of results; it does not establish correctness over every possible schedule. Polymorphism and virtual inheritance came up, along with another DSA question that was not named.
Round 4 — previous work and fundamentals
This covered past work, virtual memory, multithreading and design patterns; no particular pattern or design target was shared.
After more than three weeks under review, the candidate thought they had been rejected. They had not received a confirmed rejection when they wrote the post.
Firmware Development Engineer
1 account in this role.
What came up
Reported topics: coding tasks or technical assessments; C/C++ or memory questions.
05: Firmware: live C++ coding in Notepad despite the earlier guidance
Report date: January 23, 2026 · Firmware Development Engineer · Rejected
The commenter said both interviewers asked for C++ bit-manipulation code over screen share, using Notepad. That differed from the recruiter's reported guidance that there would be no live coding.
The same author later clarified that they had forgotten the exact operations. Those follow-ups belong to this one application. The thread starter's AI/ML team was not independently established as this commenter's team.
Software Validation and Debug Engineer
1 account in this role.
What came up
Reported topics: concrete design task.
Questions we found
Low-Level Design (LLD/OOD)
| Question | Frequency | All last seen dates |
|---|---|---|
| Design a random card shuffler and its tests | 1 | Reports: 2025-10 (interview) |
06: Validation and debug: shuffle a deck, then test it
Interview: October 2025 · Software Validation and Debug Engineer · Markham, Canada · No offer
This adjacent validation role had a roughly 50-minute technical interview. The concrete task was to design a system that shuffles cards randomly and explain how to test it. The candidate did not provide the API or a detailed randomness specification. The linked standard-deck exercise states those assumptions explicitly.
Role not reported — GPU telemetry design
1 account in this role.
What came up
One Zoom screen centered on collecting GPU performance data, with scaling and failure follow-ups.
Questions we found
System Design
| Question | Frequency | All last seen dates |
|---|---|---|
| Design a system to collect GPU performance data | 1 report | Reports: 2025-11-02 (report); Jul–Sep 2025 (header interview period) |
07: A conversational Zoom design interview about GPU performance data
Interview period: Jul–Sep 2025 (header) · Report: Nov 2, 2025 · Role and hiring level not reported · Zoom
The interviewer asked for a GPU performance-data collection system. The candidate explained an approach and answered follow-ups about scaling it and handling a component failure.
Kafka, Flink and a time-series database came up as possible choices. The interviewer focused on the reasons for choosing components and on trade-offs, rather than requiring one fixed stack or a completely polished architecture. The report supplies no scale figures or exact retention schedule.
The interview involved screen sharing and conversation, with no coding or puzzles. It ended with a few behavioral questions whose wording was not retained, then questions from the candidate about the team. The candidate found the exchange easier and more conversational than expected. Their hiring level and exact job title were not established.
C++ Developer
1 account in this role.
What came up
Reported topics: coding tasks or technical assessments; C/C++ or memory questions; projects or previous work; OS or concurrency; database or retrieval questions.
Questions we found
Algorithms / Data Structures
| Question | Frequency | All last seen dates |
|---|---|---|
| Coin Change in the C++ loop (variant not specified) | 1 | Reports: 2025-06-12 (report); Additional same-company practice history: 2026-09-06 |
| Stock profit when the sale must precede the buy | 1 | Reports: 2025-06-12 (report) |
08: C++ developer: Coin Change and selling before buying
Report date: June 12, 2025 · C++ Developer · Not selected in the final round
Round 1 — DSA, 60 minutes
The candidate reported an array-logic task, a string task and Coin Change. Only the last was named; the source does not distinguish minimum coins from counting ways.
The linked Minimum Coins for an Amount uses an explicitly authored minimum-coins variant for practice.
Round 2 — technical, 60 minutes
They discussed abstraction, inheritance, polymorphism, multithreading and concurrency. SQL exercises involved loops, conditions, subqueries and functions, but their actual retrieval goals were not preserved.
The stock problem reversed the usual order: sell first, then buy. The sell-before-buy practice task keeps that order; its one-transaction and boundary rules are practice assumptions.
Round 3 — fundamentals and projects, 60 minutes
OS scheduling, memory and threads were followed by network protocols/layers, real-world OOP modeling, and past work, ownership and design choices. No domain was given for the modeling discussion.
Software Engineer
1 account in this role.
What came up
Reported topics: coding tasks or technical assessments.
09: Germany: eight interviews, four of them technical assessments
Interviews: 2022 · Account posted August 9, 2023 · Software Engineer · Germany · Offer declined
This candidate reported eight interviews, including four technical assessments or quizzes, and ultimately declined the offer. They did not describe the individual tasks.
The comment was posted in 2023 and explicitly referred to the previous year's hiring process.
Comments
0No comments yet — be the first to share what you know.