ping·aiAsk about courses

Learning path for AI-native builders

Install the tools.
Apply them to business.

Courses taught at CSTU — from Multi-Agent Systems with CrewAI to Agentic AI with OpenClaw. Go straight into 7 Figures of Impact to build a business-application workflow — environment setup is included — or start with the $980 foundation lab first if you'd rather ease into the tools at a slower pace.

Current offerings

Two ways to start.

4 weeks · 12 hands-on hours

$980

AI Builder Foundation Lab

A 4-week foundation course for knowledge workers — founders, leaders, and ICs alike: install the AI toolchain, learn the basic concepts behind agentic workflows, and complete a small business-facing project. A standalone course — optional if you'd rather ease into the toolchain slowly.

You will ship

  • A working AI toolchain
  • Core AI-builder concepts
  • A business-facing practice project
View courseAsk about the course

13 sessions · 22.5 hands-on hours

$1,470

7 Figures of Impact

A 13-session, 1.5 CSTU semester-credit-hour course for PMs, transformation leads, and Chiefs of Staff: build the personalized-outreach workflow — in Claude Code — that produced seven figures in annual savings, proven by a 10%→70% response-rate lift. Fully self-contained, no prerequisite.

You will ship

  • A working outreach agent
  • A measured response-rate lift
  • A reusable Skill for your team
View courseRegister for - Builder With Agentic AI Cohort by 8/31/2026

CSTU graduate courses taught by Ping Wu

Past course offerings

Each course evolved from the last: CSE/MB590 established the multi-agent foundation, CSE/MB651 connected it to real software delivery, and CSE 642 turned the workflow into an agent-operating system that graduates own and operate.

CSE 642

Agentic AI and Autonomous Systems

May 2026

Trace the evolution from machine learning to agent harness. Build with OpenClaw (agent runtime) and OpenCode (coding agents). Produce a career infrastructure — resume, GitHub portfolio, LinkedIn — powered by agents you operate. The compass course for CSTU's ETTP program.

Key topics

  • OpenClaw agent runtime: sessions, memory, skills, cron, Slack integration
  • OpenCode agents: configuration, MCP, Skills, sub-agents
  • Docker container isolation and agent sandboxing
  • Karpathy LLM Wiki pattern for personal knowledge management
  • Agent DevOps: gateway mode, remote execution, secrets, SSL
  • WRITITATION methodology for agent instruction engineering
  • Career infrastructure: job search automation, application tracking, mock interviews

Outcome

A resume or portfolio ready for discovery + a deployed Daily News Briefing Agent

Students

CSTU ETTP program

★ Student spotlight — shipped product

Ultra Stock Analysis Pro

Mark DePledge · DePledge Technologies Inc.

A production stock analysis platform — AI-powered technicals, fundamentals, and grounded market sentiment — built and deployed live in a single trimester. Not a class exercise: a shipped product with a company behind it.

See it live ↗

CSE/MB 651

Software Development with Agentic AI

January 2026

Design, develop, and deploy software with an AI-assisted coding workflow. Serves as the transition from Python fundamentals to advanced fields. Covers the full stack from environment setup to production operations.

Key topics

  • AI coding agents: Gemini CLI, Claude Code, GitHub Copilot, Codex CLI
  • Software environment: Docker Desktop, command line, IDE configuration
  • Databases and storage: SQL, NoSQL, file formats (JSON, YAML, XML)
  • Docker containers and Docker Compose
  • Production operations: DNS, SSL, secrets management, release checklists
  • AI-native frameworks: CrewAI, LangChain, Google ADK, AutoGen
  • Vibe coding methodology: PRD docs, behavior-driven development

Outcome

A professional GitHub repository with complete documentation and portfolio presentation

Students

CSTU graduate program

CSE/MB 590

Multi-Agent Systems (MAS) with CrewAI

November 2024 · May 2025

Transform aspiring AI developers into confident builders who can create, customize, and deploy production-ready multi-agent systems without platform lock-in. Covers CrewAI framework in depth, agent communication patterns, and AI operations.

Key topics

  • Multi-agent architecture patterns and enterprise use cases
  • CrewAI: agent design, role definition, inter-agent communication, task decomposition
  • Memory systems, context preservation, custom tools
  • MLOps: Docker containerization, CI/CD pipelines, model versioning
  • MCP (Model Context Protocol) for tool connectivity
  • Alternative frameworks: LangGraph, AutoGen, Atomic Agents
  • Production deployment and monitoring (Prometheus, Grafana)

Outcome

Three multi-agent systems + MLOps pipeline + professional portfolio

Students

22 students (first cohort)

Real outcomes from real students

Student success stories.

MD

Mark DePledge

CSE 642 · May 2026

Built and deployed a production stock analysis platform with AI-powered technical, fundamental, and sentiment analysis — from template to live site in a single trimester. Uses advanced multi-agent workflows for market scanning, earnings analysis, and institutional-grade reporting.

Shipped product

Ultra Stock Analysis Pro↗

AI-assisted stock analysis platform with real-time technicals, fundamentals, and grounded market sentiment. DePledge Technologies Inc.

?

Your name here

Join the next cohort

Be the next success story. Clone a template, customize it with AI assistance, and ship a landing page for your own service offer.

Get started

AI Builder Foundation Lab↗

The next cohort is open. Build, ship, and list in the CSTU marketplace.

AI Harness strategy

Cost-effective AI tools.

Courses use open, affordable models via OpenRouter with OpenCode and Hermes CLIs as the primary agent harness. You learn the patterns first, then adapt to whichever models work best for your use case.

Primary agent harness

OpenCode + Hermes 4 70B

$0.13/$0.40 per 1M tokens — advanced agentic capabilities, excellent for multi-turn coding workflows. Routes to multiple providers via OpenRouter.

Heavy coding

GLM 5.2 (Z.ai) via OpenRouter

$0.94/$3.00 per 1M tokens — 1M context window, strong coding and tool use for complex refactoring and deployment tasks.

Fast / predictable

GPT-4o-mini

$0.15/$0.60 per 1M tokens — OpenAI's most capable small model, reliable routing, good for quick tasks and debugging.

OpenRouter integration

OpenRouter API is OpenAI-compatible. Swap the model slug in the code and keep the rest of your stack unchanged. Both OpenCode and Hermes CLI support custom model endpoints.

Start with the current course

Get ready for AI business applications.

Join AI Builder Foundation Lab. In 4 weeks, you'll build a personalized landing page and prepare it for listing in the CSTU online service offering network.

Start with the foundation