LLD Topic Deep-Dive Packs
Topic-first practice material derived from the master MangaAssist LLD question bank.
How to Use These Files
- Start with one topic and answer the "Primary prompt" out loud before reading the follow-ups.
- After each answer, use the staged follow-up chain to simulate a tougher interviewer.
- Only read the "Strong answer markers" after you have attempted the scenario.
- Cycle through the topics in order for breadth, or jump directly to your weakest area.
Topic Map
For a full bidirectional mapping to AWS AIP-C01 Domain 2 skills, see ../skill-crossref.md.
| # | File | Questions | Domain 2 Skills |
|---|---|---|---|
| 1 | 01-orchestrator-request-flow.md | Q1, Q11, Q24, Q38, Q46 | 2.1.1 · 2.1.2 · 2.1.3 · 2.4.1 · 2.4.3 |
| 2 | 02-intent-classification-entity-resolution.md | Q6, Q7, Q21, Q31, Q32 | 2.1.2 · 2.1.4 |
| 3 | 03-conversation-memory-session-storage.md | Q5, Q8, Q9, Q16, Q22, Q28, Q35, Q36, Q39 | 2.1.1 · 2.3.1 |
| 4 | 04-rag-indexing-retrieval-re-ranking.md | Q12, Q13, Q14, Q15, Q23, Q26, Q30, Q34 | 2.1.6 · 2.1.7 · 2.5.5 |
| 5 | 05-guardrails-validation-safety.md | Q17, Q19, Q20, Q27, Q33, Q42, Q48 | 2.1.3 · 2.3.3 |
| 6 | 06-api-contracts-streaming-and-escalation.md | Q2, Q3, Q4, Q10, Q18, Q25, Q29, Q44, Q45 | 2.1.5 · 2.4.2 · 2.5.1 |
| 7 | 07-analytics-observability-and-feedback-loops.md | Q24, Q37, Q40, Q47, Q49 | 2.3.2 · 2.5.6 |
| 8 | 08-scale-failures-and-architecture-evolution.md | Q31, Q41, Q43, Q45, Q48, Q50 | 2.2.1 · 2.3.4 · 2.4.3 · 2.4.4 |
Focus areas by file: 1. Control flow, latency budgets, retries, timeouts, growth path of the orchestrator. 2. Two-stage classification, confidence thresholds, ambiguity handling, model drift. 3. Session schema, summarization, TTL, item-size limits, hot partitions, branching. 4. Chunking, embeddings, HNSW, reranking, contradictions, freshness pipeline. 5. Safety sequencing, ASIN and price validation, PII scrubbing, testing, audit loops. 6. Request and response contracts, endpoint boundaries, WebSocket behavior, multilingual support. 7. Event evolution, prompt A/B tests, dashboards, metrics, traces, alarms. 8. Failure drills, redundancy, model swaps, multilingual evolution, launch review.
Recommended Practice Order
- Orchestrator and request flow.
- Intent classification and ambiguity.
- Memory and session storage.
- RAG and retrieval quality.
- Guardrails and validation.
- API contracts and streaming.
- Analytics and observability.
- Scale, failures, and architecture evolution.
Interview Loop
- Round 1: Give a clean two-minute answer.
- Round 2: Answer the first follow-up with concrete thresholds, timeouts, and storage limits.
- Round 3: Answer the second follow-up with failure handling and fallback behavior.
- Round 4: Answer the third follow-up with metrics, trade-offs, and operational impact.
If you want a live mock interview, open any topic file and stop at the first scenario. Do not read ahead.