Become an Engineer
Who Designs AI

Agentic Architecture

Design autonomous AI systems with Human-on-the-Loop governance and Governance-as-Code.

Harness Engineering

Control non-deterministic agents with deterministic pipelines using the Ralph Loop.

Multi-Agent SDLC

Build autonomous development pipelines where Planner, Coder, and QA agents collaborate.

Open-Source MLOps

Deploy and operate DeepSeek-Coder-V2 on DGX H100 + vLLM.

Course Roadmap

Phase 1: Agentic System Foundations

• Week 1 AI Systems Paradigm Shift
• Week 2 HOTL Governance & Governance-as-Code
• Week 3 MCP Architecture & Agentic Tool Ecosystem

Phase 2: Harness Engineering

• Week 4 Loop Paradigm — Iteration Beats Complexity
• Week 5 Context Management & Preventing Context Rot
• Week 6 Instruction Tuning

Phase 3: Multi-Agent SDLC

• Week 7 Multi-Agent SDLC Design
• Week 8 Project Proposal Presentation (Midterm Replacement)
• Week 9 Planner & QA Agent Implementation

Phase 4: Open-Source Models & MLOps

• Week 10 Open-Source Coding LLMs & Local Deployment
• Week 11 vLLM High-Throughput Inference Optimization
• Week 12 Telemetry & LLM-as-Judge

Phase 5: Capstone Ralphthon

• Week 13 Capstone Project Kickoff
• Week 14 Ralphthon Execution
• Week 15 System Integration & Final Testing
• Week 16 Final Presentations & Wrap-up

Quick Start

Syllabus Course objectives, grading criteria, and weekly schedule

Week 1: AI Systems Paradigm Shift From AI models to AI systems — HITL/HOTL/HIC governance and the dawn of the agentic era

Lab 01: Environment Setup Development environment, DGX access, AI coding CLI installation

Contributing Found an error? Send a PR to improve the course

---

Cheju Halla University, Dept. of AI · Spring 2026 · Instructor: Young Joon Lee