CSE 598 – Agentic AI (Fall 2026 @ASU)
Course Number: CSE 598
Faculty Instructor: Hua Wei, Ph.D.
E-mail: hua.wei AT asu.edu
Overview
This course explores how large language models can be used to build autonomous agents capable of reasoning, planning, using tools, and interacting with complex environments. Students will study core topics including agent architectures and the agent loop, planning and reasoning, tool use and function calling, agent evaluation and benchmarks, world models and memory, instruction following and policy learning, context engineering and retrieval-augmented generation (RAG), multi-agent collaboration, and the critical issues of safety, trustworthiness, and privacy in agentic systems. A distinctive feature of this course is a parallel source-reading track: every concept we study in lecture is then located and examined in the source code of OpenAI's open-source Codex CLI. The course culminates in a team capstone modeled on the Forward Deployed Engineer (FDE) practice: students discover a real problem with a stakeholder, prototype and evaluate against a baseline, harden and deploy, and iterate on feedback.
Prerequisites
Students are expected to have graduate standing, proficiency in Python, and prior coursework or experience in machine learning or NLP. A working knowledge of probability, linear algebra, and the academic research process is assumed. Familiarity with reading source code is expected, and Rust experience is helpful but not required.
Textbook
There is no required textbook. Below are some recommended reference books.
- Antonio Gulli, Agentic Design Patterns: A Hands-On Guide to Building Intelligent Systems, Springer, 2025 (with accompanying runnable notebooks). (book website).
Course Structure
Lectures will consist of instructor-led presentations and demos on course topics, along with short quizzes on the lectures and assigned readings. Labs will consist of student-led code-reading presentations and paper sharing, guided navigation of the Codex source code, and time for team capstone work and Q&A with the TA. Students are expected to complete their source-reading reports with a few additional hours outside of class. This means that students are required to bring their laptops to class and labs and ensure they are fully charged and ready to complete assigned work.
Tentative Topics
Week | Unit |
|---|
Week 1 – 2 | Foundations of LLM-Based Agents What makes a system "agentic": MDP formalization, agent architectures, and the agent loop |
Week 3 – 5 | Core Agent Capabilities and Evaluation Planning and reasoning, tool use and function calling, evaluation, and benchmarks |
Week 6 – 7 | State and Memory World models, state representation, memory, and long-term context |
Week 8 – 10 | Context Engineering and Learning Prompt design, knowledge retrieval / RAG, instruction following, and policy learning (RLHF, DPO/GRPO) |
Week 11 – 12 | Multi-Agent Systems and Safety Coordination and communication; trustworthiness, prompt injection, guardrails, human-in-the-loop |
Week 13 | Advanced topic: Applications Deep Dive |
Week 14 | Future Directions & Demo Day |
Assignments and Grading
Component | Weight | Details |
|---|
Participation and in-class quizzes | 10% | attendance, discussion, peer review, and short quizzes on the lectures. |
Codex source-reading reports | 30% | short write-ups locating and explaining where a course concept is implemented in the OpenAI/Codex source. All reports carry equal weight. A missed report receives a zero. |
Paper Sharing | 10% | Students are organized into paper-sharing groups of five, separate from their capstone teams. Each group leads one paper sharing during lab, presenting an assigned reading and facilitating discussion. |
Project (Forward Deployed Engineer model) | 50% | Students take a real stakeholder’s problem, prototype against real data, deploy a reliable agentic system, and iterate on feedback. |
CSE 572 – Data Mining (Fall 2023- Fall 2026 @ASU)
Course Number: CSE 572
Faculty Instructor: Hua Wei, Ph.D.
E-mail: hua.wei AT asu.edu
Overview
This course will introduce fundamental concepts and techniques in data mining including classification, clustering, dimensionality reduction, and outlier detection. Students will learn the theory behind topics as well as gain hands-on experience implementing data mining techniques and applying them to real-world problems and data. Students will learn how data mining is used in research and gain an understanding and practice of the complete research process.
Prerequisites
- Students are expected to have a working knowledge of basic probability theory, linear algebra, and the academic research process.
- Note: Assignments and projects should be implemented in Python.
Textbook
There is no required textbook. Below are some recommended reference books.
- Jiawei Han, Micheline Kamber and Jian Pei, Data Mining: Concepts and Techniques, 3rd ed.
- Chris Bishop, Pattern Recognition and Machine Learning.
- Pang-Ning Tan, Michael Steinbach, Anuj Karpatne, and Vipin Kumar, Introduction to Data Mining, 2nd ed.
Assignments and Grading
| Participation | 5% |
| In class lab assignment | 20% |
| Assignment, quiz | 35% (5% quiz, 10% assignment 1, 10% assignment 2, 10% assignment 3) |
| Project | 40% (5% proposal, 5% literature review, 5% progress report, 15% final presentation, 10% final report) |
- Assignments: All the assignments are done individually.
- Project: The course project is carried out as a team.
- Class attendance: Attending class is required. Excused absences should get approved by the instructor BEFORE the class.
IS392 – Web Mining (Spring 2022/2023 @NJIT)
Course Number: IS392-002
Classroom: Tiernan Hall 113 (after Jan. 30)
Class Meets: 11:30 am – 12:50 pm, Monday & Wednesday,
Faculty Instructor: Hua Wei, Ph.D.
E-mail: hua.wei AT njit.edu
Office: GITC 3803H
Office Hours: Monday 1-2 pm, or by appointment
Overview
This course introduces the design, implementation, and evaluation of web mining applications. Topics include automatic indexing, natural language processing, retrieval algorithms, basic machine learning techniques, and their applications to web data. Students will gain hands-on experience applying theories in case studies.
Prerequisites
- IS218 OR IT114 OR CS114
- Programming, linear algebra, probability, algorithm analysis, data structure.
- Note: Assignments and projects should be implemented in Python.
Textbook
There is no required textbook. Below are some recommended reference books.
- Search Engines: Information Retrieval in Practice, by Croft, Metzler, and Strohman. Publisher: Addision-Wesley
- Paper 1: What Do People from Information Retrieval?, W. Bruce Croft
- Paper 2: Search Engine Optimization Starter Guide, Google
- Jiawei Han, Micheline Kamber and Jian Pei, Data Mining: Concepts and Techniques, 3rd ed.
- Kevin Patrick Murphy, Machine Learning: a Probabilistic Perspective, 2012./li>
Assignments and Grading
| Assignment, quiz | 45% (5% quiz, 14% assignment 1, 13% assignment 2, 13% assignment 3) |
| Project | 45% (10% report1, 10% report2, 10% report 3, 15% final report) |
| Class Attendance | 10% |
- Assignments: All the assignments are done individually.
- Project: The course project is carried as a team.
- Class attendance: Attending class is required. Excused absence should get approved by the instructor BEFORE the class.
IS 657 Spatiotemporal Urban Analytics (Fall 2022 @NJIT)
Course Number: IS657
Classroom: Jersey City 101
Class Meets: Tuesday from 6:00 – 8:50 pm,
Faculty Instructor: Hua Wei, Ph.D.
E-mail: hua.wei AT njit.edu
Office: Faculty office @JerseyCity
Office Hours: Tuesday 4 – 6 pm, or by appointment
Overview
Cities now generate an immense amount of publicly accessible data that allows us to ask and answer new questions about cities and urban populations. This course will teach the methods, models, and tools for data-driven urban research. You will learn the basics of urban data acquisition, ethics, management, visualization, and statistical analysis with a focus on spatio-temporal data. By the end of the course, you will be able to formulate a question relevant to urban science and then acquire, prepare, and analyze data to gain insights and aid in decision making. The class format will include lectures from the professor, labs using Python, readings, discussion, and student projects.
Prerequisites
- IS-665
- Basic Python programming knowledge (Required since language instruction will not be covered in class)
Textbook
There is no required textbook. Below are some recommended reference books.
- Singleton, Spielman, and Folch. “Urban Analytics”
- Zheng. “Urban Computing”
- Jiawei Han, Micheline Kamber and Jian Pei, Data Mining: Concepts and Techniques, 3rd ed.
Assignments and Grading
| Assignment, quiz | 40% |
| Project | 50% |
| Class Attendance | 10% |
- Assignments: All the assignments are done individually.
- Project: The course project is carried out as a team.
- Class attendance: Attending class is required. Excused absences should get approved by the instructor BEFORE the class.
