▸case-01 I claim that Raft guarantees leader election in a single round if all nodes are healthy. I am preparing for a senior infrastructure engineer interview. Test my understanding by challenging my premise with a edge case scenario. | fail→pass | 10,924 | 10,240 | -6% | 1 | 1 | 0% | 1,840 | 1,854 | +1% | 0 | 0 | — |
▸case-02 I believe that putting column A or column B first in a PostgreSQL composite B-tree index (A, B) makes no performance difference for queries filtering on both. Drill down into my understanding with a targeted follow-up question. | fail→fail | 7,104 | 8,448 | +19% | 1 | 1 | 0% | 1,327 | 1,486 | +12% | 0 | 0 | — |
▸case-03 I told my interviewer that SYN cookies prevent SYN flood attacks by storing incoming connection states in kernel memory buffers. Challenge my claim directly and ask me to explain how state is actually handled. | pass→pass | 5,020 | 3,459 | -31% | 1 | 1 | 0% | 948 | 567 | -40% | 0 | 0 | — |
▸case-04 I insist that using Python multi-threading is always faster than multi-processing for CPU-bound matrix multiplication tasks because threads share memory space. Put me on the spot with a probe about how execution lock scheduling affects CPU-bound thread performance. | pass→fail | 11,405 | 5,389 | -53% | 1 | 1 | 0% | 1,544 | 805 | -48% | 0 | 0 | — |
▸case-05 I claim 2PC (Two-Phase Commit) is fully fault-tolerant because if a participant crashes during the prepare phase, the coordinator simply proceeds to commit. Interrogate my knowledge on coordinator and participant failure modes. | pass→pass | 10,168 | 5,986 | -41% | 1 | 1 | 0% | 1,857 | 898 | -52% | 0 | 0 | — |
▸case-06 I stated in my mock interview that returning a pointer to a local stack variable in C is completely safe as long as the calling function reads it immediately. Interrogate my understanding of stack frame lifecycles. | pass→fail | 10,840 | 7,122 | -34% | 1 | 1 | 0% | 1,932 | 986 | -49% | 0 | 0 | — |
▸case-07 I told my mock interviewer that using array index as the React key prop in a dynamic sorting list is fine because React uses key array positions to track element identity. Challenge my assumption with a dynamic reordering scenario. | pass→pass | 12,906 | 6,746 | -48% | 1 | 1 | 0% | 2,629 | 1,139 | -57% | 0 | 0 | — |
▸case-08 I claim Apache Kafka guarantees total strict message ordering across all partitions in a topic regardless of partition count. Test my system design knowledge by asking me to correct this assertion. | pass→pass | 6,762 | 4,300 | -36% | 1 | 1 | 0% | 1,027 | 684 | -33% | 0 | 0 | — |
▸case-09 I put 'COPY . .' before 'RUN npm install' in my Dockerfile because it keeps the build setup clean. Test my understanding of layer caching performance impact during frequent code changes. | pass→pass | 11,625 | 5,571 | -52% | 1 | 1 | 0% | 2,210 | 904 | -59% | 0 | 0 | — |
▸case-10 I asserted that GraphQL natively eliminates the N+1 database query problem simply by allowing nested field selection in single queries. Interrogate my understanding of backend field resolver execution mechanics. | pass→pass | 14,478 | 5,504 | -62% | 1 | 1 | 0% | 2,417 | 854 | -65% | 0 | 0 | — |
▸case-11 I argued that TLS 1.3 session resumption using PSK (Pre-Shared Keys) provides perfect forward secrecy even if the main server long-term key is compromised retroactively. Challenge my security model understanding. | pass→pass | 16,370 | 10,792 | -34% | 1 | 1 | 0% | 2,962 | 1,829 | -38% | 0 | 0 | — |
▸case-12 I claim that setting short identical TTLs across all high-frequency cached items is the best way to keep Redis memory usage low and prevent backend load spikes. Probe my understanding regarding concurrent cache expiration behavior. | pass→pass | 10,454 | 4,193 | -60% | 1 | 1 | 0% | 1,794 | 639 | -64% | 0 | 0 | — |
▸case-13 I claim that setting a liveness probe to ping an external dependent database endpoint ensures high availability of my pod. Challenge my understanding of container lifecycle recovery. | pass→pass | 14,844 | 5,458 | -63% | 1 | 1 | 0% | 2,066 | 890 | -57% | 0 | 0 | — |
▸case-14 I asserted during a system design interview that a distributed system can achieve simultaneously high Consistency, Availability, and Partition Tolerance during a network split if tuned correctly. Push back on this claim. | pass→pass | 13,628 | 15,771 | +16% | 1 | 1 | 0% | 2,444 | 2,205 | -10% | 0 | 0 | — |
▸case-15 I claim generational garbage collectors divide heap memory into young and old generations because older objects die faster than newly allocated objects. Grill me on the underlying empirical memory allocation patterns. | pass→fail | 12,885 | 8,047 | -38% | 1 | 1 | 0% | 1,745 | 1,291 | -26% | 0 | 0 | — |
▸case-16 I believe PKCE in OAuth 2.0 is redundant for public native mobile applications if HTTPS client transport is enforced. Ask a targeted counter-question regarding local application scheme redirection risks. | fail→fail | 5,455 | 5,435 | -0% | 1 | 1 | 0% | 791 | 722 | -9% | 0 | 0 | — |
▸case-17 I told an interviewer that in CQRS architecture, read model updates from the event store are strictly synchronous and immediate, ensuring zero read-model stale data. Test my understanding of event-driven read synchronization. | fail→fail | 9,506 | 9,530 | +0% | 1 | 1 | 0% | 1,701 | 1,214 | -29% | 0 | 0 | — |
▸case-18 I stated that HTTP/2 completely eliminates all forms of head-of-line blocking across the entire network stack because it multiplexes streams over one TCP connection. Interrogate my transport protocol knowledge. | pass→pass | 10,209 | 3,564 | -65% | 1 | 1 | 0% | 1,697 | 523 | -69% | 0 | 0 | — |
▸case-19 I argued that SQL REPEATABLE READ isolation level strictly prevents phantom reads across all relational database implementations. Challenge me on vendor-specific isolation behavior differences. | pass→fail | 14,355 | 7,503 | -48% | 1 | 1 | 0% | 2,711 | 1,279 | -53% | 0 | 0 | — |
▸case-20 Here is a resume bullet point for a backend developer: 'Worked on database queries and improved speed.' Rewrite this into a high-impact, STAR-formatted resume bullet point with metric placeholders. | pass→fail | 7,625 | 8,650 | +13% | 1 | 1 | 0% | 1,317 | 1,599 | +21% | 0 | 0 | — |
▸case-21 I am a high school student learning to code. Explain what an API is in simple, friendly terms using a restaurant analogy. | pass→fail | 8,667 | 9,487 | +9% | 1 | 1 | 0% | 1,521 | 1,708 | +12% | 0 | 0 | — |
▸case-22 Refactor this JavaScript function to use async/await syntax instead of raw promises:
function fetchData() { return fetch('/api/data').then(res => res.json()); } | fail→pass | 4,134 | 5,651 | +37% | 1 | 1 | 0% | 856 | 995 | +16% | 0 | 0 | — |