Evidence pack và interview narrative end-to-end
Demo đáng tin phải reproducible, time-boxed và cho thấy một failure/recovery; slide architecture không thay thế executable evidence.
Evidence pack
README one-command, requirements/invariants, C4/sequence/state diagrams, ADRs, API/event schemas, test/coverage reports, pipeline/SBOM, dashboards/traces, load/failure results, runbooks/postmortem.
Ten-minute narrative
1 phút problem/constraints; 2 architecture/data flow; 2 invariants/failures; 2 security/operations; 2 demo evidence; 1 trade-offs/next steps. Luôn bắt đầu từ requirement.
Mock loop
45–60 phút: walkthrough, deep dive, incident/debug scenario, change requirement và behavioral ownership. Interviewer hỏi alternatives, scale limit, rollback và proof—not trivia.
Iteration
Record questions chưa trả lời, unsupported claims, demo fragility và time overruns. Mỗi gap thành issue có priority; không thêm technology nếu chưa cải thiện requirement/risk/evidence.
English opening
This project models an order and payment workflow where duplicate requests, partial failures, and unknown outcomes are first-class concerns. Each service owns its data. We begin with a synchronous flow for simplicity, then evolve to an outbox-based asynchronous flow while preserving idempotency and auditability.
Interview evidence package
architecture/: context/container/sequence/failure diagrams.decisions/: ADRs với alternatives và revisit triggers.runbooks/: timeout, DLQ, stuck payment, rollback, restore.evidence/: test reports, load results, trace screenshots/exports và incident timeline.demo-script.md: 10 phút happy path + failure path bằng Việt/Anh.
Mock interview schedule
- Round 1 (60m): Java/Spring/testing/database live questions.
- Round 2 (45m): API/security/frontend.
- Round 3 (60m): distributed systems + messaging failure walkthrough.
- Round 4 (60m): system design based on capstone at 10x/100x scale.
- Round 5 (45m): Kubernetes/CI/CD/observability incident.
- Round 6 (45m): behavioral/leadership in English.
Scoring: correctness, depth, trade-off, failure reasoning, testing evidence, security, data consistency, observability, communication VI và communication EN; mỗi category 0–2, target 16/20 và không category nào bằng 0.