Pandai Open — Free, async, anyone
Twelve weeks. Three phases.Ship something useful.
The same curriculum we teach in-person. Watch any module. Ship the SHIP project. If you’re shipping by Module 6, you’re our kind of applicant for Cohort 2.
Aligned with Komdigi DTS & Kepmen Kominfo 117/2024.
Why Pandai Open
A free curriculum, on purpose.
Pandai Open is the brand engine and the funnel. The cohort is what gets you placed; Pandai Open is what proves you can ship.
Free, no login.
Watch on YouTube. Code along. No signup wall. No "premium tier" upsell.
Same content as Cohort 1.
The cohort buys you operational simulation and placement, not better videos. The curriculum is identical.
A real funnel.
Top performers in Pandai Open get fast-tracked into Cohort 2 admissions. Ship work; we’re watching.
How each module works
Curated external fundamentals + the Pandai layer.
We don’t teach fundamentals. Each module starts with curated YouTube videos from the people who built this stuff — Karpathy, the Anthropic team, others. Then Pandai layers on the SHIP project, Indonesian context, and the AI-Native discipline.
External videos + Pandai SHIP project + AI-Native code review
Watch the intro — free, no login
Two moves in AI engineering. Start with the first.
Three public intros to set the frame. The remaining nine modules live at app.pandai.academy — free, sign in once with Google.
01 / 12 · Using AI. Then bending it.
02 / 12 · Under the hood.
Open the deep dive →03 / 12 · Applications.
Open the module →The full curriculum
Twelve weeks. Three phases. One arc.
Vibe coding first, engineering depth second, track specialization third. You ship something useful with Claude in Week 1; by Week 4 you’ve built an agent that does real work; by Week 12 you’re a specialist with a real-client capstone.
Phase 1 · Pre-week → Week 4
AI Vibe Coding
Ship something useful with Claude. By Friday of Week 4 you’ve built an agent that does real work — not a toy.
- W1
Claude Code as your dev environment
Set up your coding agent — Claude Code by default. Use AI to automate something you actually do every week. By Friday: SHIP a CLI that handles a real annoyance in your daily workflow. (No MCP-writing yet — just real use.)
- W2
LLM fundamentals — RLM first
Now that you’ve used AI for a week, look under the hood. What an LLM actually is. The vocabulary: agent vs agentic-AI, automation vs conversation, RLM vs RAG, MCP. Why long context beats retrieval until it doesn’t — no RAG yet.
- W3
Agents and MCP
The full arc from automation to agentic AI: conversation → tool use → multi-agent → RAG → MCP, as the map of the territory. Use off-the-shelf agents and MCPs together — assemble a 3-tool agent from existing pieces. (Writing your own MCP comes in W4.)
- W4
Practical agent shipping
Pick one and SHIP it: video, slides, Excel, Indonesian forms (KTP / NPWP / tax), or website registration. Real artifact, real user, real Friday.
Phase 2 · Week 5 → Week 7
Engineering Depth
From toy to product. From solo coder to AI-orchestrator. End-to-end LLM apps with auth, evals, observability, and the manager mindset that ships them.
- W5
Bella case study
End-to-end LLM product, opened up: auth, schema, eval harness, observability. Prompt patterns that survive production — including the 100% / 40% memory rule. The Bella architecture, fully exposed.
- W6
Become the CEO of your codebase
Graduate from solo coder to AI-orchestrator. Run the Think → Plan → Build → Review → Ship → Reflect loop with Claude Code subagents playing roles you used to do alone. First principle: agents are non-deterministic — so you ship observability, not certainty. Cohort students: Gate 1 begins for top 30% of placements.
- W7
Pick your track
Five tracks, one day each: AI Engineer, Software, Data, Quality, Tech Ops. Build a small artifact in every track. Friday: rank what you want for the last five weeks.
Phase 3 · Week 8 → Week 12
Track Specialization
Pick your specialty. Spend the last five weeks going deep on one of five tracks — AI Engineer, Software, Data, Quality, or Tech Ops.
- W8
Track depth begins — evals before everything
Tracks branch from here. AI Engineer goes evals-first — the discipline that separates demos from products you can sell. Software, Data, Quality, Tech Ops each enter their own depth.
- W9
Build your first closed-loop system
AI Engineer: SHIP an LLM-wiki — a system that researches questions, grades its own answers, and updates what it knows. The ontology you build defines what good means to the agent. Cohort students: Gate 2 begins for the next 40% of placements.
- W10
RAG, finally — for what RLM can’t carry
AI Engineer: retrieval, advanced RAG (chunking, pruning, compacting context), GraphRAG over your ontology. The cases where reasoning-with-context beats long context. Other tracks: integration patterns and capstone scoping.
- W11
Self-improving agents in production
AI Engineer: SHIP a verifiable closed-loop — your agent runs, grades its own output, and is measurably better Friday than Monday. MLOps essentials: serving, prompt versioning, token analytics, human-in-loop. Other tracks: capstone build for a real client.
- W12
Demo Day + Gate 3
Capstones presented to the cohort. AI Engineer: self-improving autoresearch demoed live — the system gets better while the audience watches. Plus a final pass on go-to-market thinking: who is this for, how do you reach them. Top performers join Metatech as FDEs.
Each row pairs a curated external video with a Pandai SHIP project. Videos deep-link to YouTube as each module’s reading list is finalised — Pandai does not produce them.
SHIP projects
Ship every Friday. Public GitHub repo.Real users.
Each module has a SHIP project. You build it, you push it, you write a Loom walking through what you built and what you got wrong. Not assignments. Real shipped artifacts.
FAQ
Asked often.
- Do I need to register for Pandai Open?
- No. Watch anything, anytime, no signup. Pandai Open is a public YouTube curriculum.
- Is Pandai Open enough to get hired?
- Maybe. The cohort gives you the operational simulation (incidents, scope changes, on-call) and the anchor BUMN placement. Pandai Open gives you the AI engineering content. Both are useful; the cohort is what gets you placed.
- Can I apply to Cohort 1 if I haven’t done Pandai Open?
- Yes. We look at what you’ve shipped, regardless of where you learned. But the more SHIP projects in your application, the stronger.
- When do videos go live?
- Production Pillar is queueing them now. M0–M3 first; you’ll see them roll out over the coming weeks. [VERIFY: timeline]