Syllabus: Enterprise Vibe Engineering

1. Program summary
Credential Certified Vibe Engineer (three levels, see section 6)
Duration 4 terms x 4 weeks = 16 weeks full programme (approx. 10 to 12 hours per week), or modules taken standalone
Format Concept lesson, AI-pairing lab, red-team review, Definition of Done gate
Prerequisite Has shipped at least one app with an AI tool (Claude Code, Cursor, Lovable, Replit, Bolt, v0 or similar). No CS degree required.
Reference stack TypeScript, Next.js, Postgres (Supabase), Redis, a queue, Docker, GitHub Actions, Cloudflare or Vercel, OpenTelemetry. Concepts are stack-agnostic.

2. Who it is for

  1. Founders and operators who build their own products with AI and are about to take real money or real data.
  2. Product people and analysts moving into building.
  3. Junior and mid engineers whose job is now reviewing AI output more than writing code.
  4. Enterprise teams rolling out AI coding tools who need a common bar for "production ready."

3. Programme learning outcomes

By graduation a student can:

  1. Specify a system clearly enough that an AI produces the right thing (specs, context, rules files, acceptance tests).
  2. Verify AI-written code against security, correctness and performance standards without taking the AI's word for it.
  3. Secure a web application against the OWASP Top 10:2025 and an AI agent against the OWASP Top 10 for Agentic Applications 2026.
  4. Design for failure: timeouts, retries, idempotency, queues, graceful degradation.
  5. Scale sensibly: know which of the 111 concepts matter now and which are premature.
  6. Ship safely: CI/CD, migrations, feature flags, one-minute rollback.
  7. Operate: SLOs, alerting, incident response, postmortems, tested backups.
  8. Govern: produce the audit trail, data-protection and AI-governance evidence an enterprise or regulator asks for (PDPA, IMDA Model AI Governance Framework for Agentic AI, SOC 2 style controls).

4. Pedagogy: the four-beat module

Every module runs the same loop.

  1. Concept (Learn): the why, the failure it prevents, a real incident.
  2. Pair (Build): student directs an AI to build the feature using provided prompts. The AI is the junior engineer; the student is the tech lead.
  3. Break (Red-team): student attacks or stresses their own build with a checklist and tooling. Most labs are designed so the naive AI output fails here on purpose.
  4. Gate (Prove): a Definition of Done with objective evidence (test output, scan report, dashboard screenshot, restore log). No evidence, no pass.

Principle: Trust, but verify is not enough for AI code. The standard is verify, then trust.

5. Course map

Term 1: Foundations (Weeks 1 to 4)

Wk Module Core question Poster items
1 00 · Orientation: The Vibe Engineer Mindset + 01 · The AI-Assisted Engineering Workflow How do I make the AI build the right thing and prove it? 6 (intro), 108
2 02 · How the Internet Actually Works What happens between the click and the database? 71, 73, 91 to 94, 97 to 99
3 03 · Data Foundations Will my data stay correct and fast at 100x? 25 to 28, 36, 37, 81, 82
4 04 · APIs and Contracts Can other systems depend on me safely? 14, 100 to 103

Term 2: Security (Weeks 5 to 8)

Wk Module Core question Poster items
5 to 6 05 · Application Security: OWASP Top 10:2025 Who can do what, and can an attacker pretend to be them? 1, 67 to 80
7 to 8 06 · AI and Agent Security What happens when the input is the attack? (beyond poster)

Term 3: Reliability and Scale (Weeks 9 to 12)

Wk Module Core question Poster items
9 07 · Resilience Patterns What does my system do when a dependency is slow or down? 10 to 13, 20, 44, 56, 57
10 08 · Async, Events and Real-Time How do I do slow work without blocking users? 15 to 19, 21 to 24, 101
11 09 · Performance and Scaling Where is the bottleneck, and what is the cheapest fix? 2 to 5, 29 to 31, 45 to 50, 88 to 90
12 10 · Distributed Systems Correctness What goes wrong when two things happen at once? 32 to 43, 95, 96

Term 4: Production Operations (Weeks 13 to 16)

Wk Module Core question Poster items
13 11 · Shipping: CI/CD, Supply Chain and Infrastructure Can I ship ten times a day and undo any of it in a minute? 6 to 9, 51 to 55, 104 to 108
14 12 · Observability and SRE How do I know it is broken before customers tell me? 56 to 66
15 13 · Disaster Recovery and Incident Management What is my plan when the worst happens? 83 to 87, 109 to 111
16 14 · Compliance, Governance and Audit + Capstone Can I prove all of this to an auditor? (beyond poster)

6. Certification levels

Level Name Requirement Signals
L1 Vibe Builder Terms 1 and 2 passed; app passes the "Do Now" section of Production Readiness Checklist Safe to take real users and data
L2 Vibe Engineer Terms 1 to 3 passed; app survives the Term 3 chaos lab Safe to take payments and run agents
L3 Enterprise Vibe Architect All terms + Capstone defended before a panel Can lead an AI-native team and pass vendor security reviews

7. Assessment and grading

Component Weight Notes
Module gates (Definition of Done evidence) 40% Binary per item; all "must" items required to pass a module
Red-team findings log 15% Quality of the vulnerabilities and failure modes students find in their own and peers' builds
Weekly architecture decision records (ADRs) 10% One ADR per module explaining a trade-off
Incident simulation (Week 15) 10% Graded on detection time, communication and postmortem
Capstone build + defence 25% See Capstone

Automatic fail conditions (any module): a secret committed to git history; a database table reachable by an unauthenticated user; an agent able to take an irreversible action without human approval.

8. Required tooling (free tiers are fine)

9. Weekly rhythm (10 to 12 hours)

Block Hours
Concept reading and lesson 2
Pair lab (build with AI) 4 to 5
Red-team session 2
Gate evidence + ADR 1 to 2
Cohort review (peer attacks your build) 1

10. Academic integrity, AI edition

Using AI is required, not cheating. What counts as misconduct here is unverified claims: submitting "tests pass" without the run log, "secure" without the scan, "backed up" without a restore.