EECS 498-016 Fall 2026

Syllabus

Three phases, 28 lectures, 12 labs, 3 projects. Dates follow the official UMich Fall 2026 calendar. Details marked TBD get locked in the week 1 syllabus.

01 At a glance

CourseEECS 498-016 · Applied Agentic Software Engineering

Credits4 · upper-level CS technical elective

TermFall 2026 · Aug 31 – Dec 11

LecturesTue + Thu · 3:00 – 4:30 PM · room TBD

LabsMon 3:30 – 5:30 or Tue 12:30 – 2:30 · attend either · room TBD

AssessmentProjects, labs, live demos · no exams

PrerequisitesEECS 281 and EECS 201 · or ULCS standing or instructor permission

LanguagePython · plus the shell and git

Contacteecs-aase-staff@umich.edu

02 Description

What this course is

AASE (say it "A two S E") teaches software engineering with AI agents in the loop. Students progress through three phases of increasing autonomy: working effectively with AI coding tools, scripting agents directly against LLM APIs, and designing a production-grade agent orchestrator as an individual capstone.

The course is relentlessly hands-on. Lectures run live demos against real repositories. Labs are working sessions with TA checkoffs. Projects ship code that runs, with evals to prove it.

By the end of the semester, students will be able to:

"Understanding is what gets graded here, not generation. If you can't explain it, you didn't build it."

/01Operate fluently in modern AI-assisted coding workflows (Aider, Claude Code, Codex)
/02Reason about AI coding failures using the Big Three: context, model, and prompt
/03Build a scripted agent from scratch using direct LLM API calls and tool use
/04Customize agentic CLIs with slash commands, hooks, sub-agents, and MCP servers
/05Design and ship a production-grade agent orchestrator

03 Weekly schedule

Fifteen weeks, calendar-anchored

Lectures meet Tuesday and Thursday, 3:00 to 4:30. Lab runs once a week with two sections covering identical content: Monday 3:30 to 5:30 or Tuesday 12:30 to 2:30, and you can go to either. Three calendar breaks shape the term: Labor Day (Mon Sep 7), fall study break (Mon–Tue Oct 19–20), and Thanksgiving (Wed–Fri Nov 25–27).

Phase 1

AI-Assisted Development

Weeks 1–3 · L01–L06 · Lab 01 · Project 1

Week	Tue lecture	Thu lecture	Lab (Mon / Tue)	Milestones
Wk 1Aug 31 – Sep 4	L01Course Intro + AI Coding Landscape	L02How LLMs Code (Under the Hood)	No lab · labs begin week 3	P1 + Aider self-study released
Wk 2Sep 7 – 11	L03The Big Three: Context, Model, Prompt	L04Spec-Driven Coding	No lab · Labor Day Mon Sep 7	Aider setup gate · Fri Sep 11
Wk 3Sep 14 – 18	L05Modern Agentic CLIs	L06From Pair-Coding to Agent Orchestration	Lab 01Aider mastery checkpoint	Project 1 due · Fri Sep 18

Phase 2

Scripted Agentic Workflows

Weeks 4–6 · L07–L12 · Labs 02–04 · Project 2

Week	Tue lecture	Thu lecture	Lab (Mon / Tue)	Milestones
Wk 4Sep 21 – 25	L07Working with LLM APIs	L08Tool Use and the Agent Loop	Lab 02Claude Code hands-on: replay a P1 task, diff against Aider
Wk 5Sep 28 – Oct 2	L09Claude Code Customization: Slash Commands, Hooks, Settings	L10Sub-agents and MCP	Lab 03Build a tool-using agent from scratch (raw API calls, ~100 lines)	P2 released · Tue Sep 29
Wk 6Oct 5 – 9	L11Programmatic Aider and the Director Pattern	L12Evals and Guardrails	Lab 04Write a custom slash command + Stop hook

Phase 3

Agent Orchestrator Capstone

Weeks 7–15 · L13–L28 · Labs 05–12 · Project 3

Week	Tue lecture	Thu lecture	Lab (Mon / Tue)	Milestones
Wk 7Oct 12 – 16	L13Orchestrator Architecture	L14State and Memory	Lab 05Orchestrator scaffold	P3 released · Tue Oct 13 Project 2 due · Fri Oct 16
Wk 8Oct 19 – 23	Fall study break · no class	L15Multi-Agent Patterns	No lab · study break Mon + Tue
Wk 9Oct 26 – 30	L16Sessions and Routing	L17MCP Deep Dive	Lab 06State + memory layer
Wk 10Nov 2 – 6	L18Tool Ecosystem and Security	L19Long-Running Tasks	Lab 07Multi-agent communication
Wk 11Nov 9 – 13	L20Milestone Selection Strategy	L21Dependency-Aware Scheduling	Lab 08MCP integration
Wk 12Nov 16 – 20	L22Evaluation Methodology	L23Comparative Analysis: Testing Agent Designs	Lab 09Eval suite
Wk 13Nov 23 – 27	L24Production Deployment	Thanksgiving · no class	Lab 10Cost optimization
Wk 14Nov 30 – Dec 4	L25Integration and Polish	L26Demo Prep	Lab 11Production deploy
Wk 15Dec 7 – 11	L27Industry Case Studies in Agentic Systems	L28Final Class: Project Showcase + Wrap-Up	Lab 12Demo dress rehearsal	P3 final demos · Dec 8 – 10

Lecture titles in the back half may shift slightly as material is finalized. Dates won't.

04 Projects

Three projects, rising autonomy

Project 1 · AI-assisted mini-project

Build a small but complete Python application using Aider as your primary tool, then complete one well-defined task in Claude Code and compare the two ways of working. Options include a CLI data dashboard, a markdown notes tool, or an API aggregator; you may propose your own with instructor approval.

ReleasedTue Sep 1

DueFri Sep 18

Effort10–12 hrs

FormatIndividual

Roughly 70% of the build through Aider, 30% through Claude Code (after L05)
Submit code, your spec, the Aider chat history, and a Claude Code session export
500-word reflection comparing the tools

Project 2 · A focused scripted agent

Pick one focused job (code reviewer, PR summarizer, test generator, doc auditor, security scanner, or similar) and build an agent that does it end to end on a real repository, calling an LLM API directly with no agent frameworks in between.

ReleasedTue Sep 29

DueFri Oct 16

Effort15–20 hrs

FormatIndividual

Calls the LLM API directly and exposes at least 2 tools to the model
Ships an eval suite with at least 5 test cases
Submit code, eval results, a design doc, and a 3-minute demo recording

Project 3 · The orchestrator

The capstone, and the largest single grade in the course. Over nine weeks, design and ship a real, deployable multi-agent system. The weekly labs walk you through scaffold, state, communication, MCP, evals, cost, and deployment so the system grows alongside the lectures.

ReleasedTue Oct 13

Final demosDec 8 – 10

Effort60–80 hrs

FormatIndividual

Multi-module, tested code repository and an architecture document
Eval suite with reproducible results, plus a cost and latency report
Observability dashboard or equivalent
Live demo in week 15 with a recorded backup

05 Grading

Weight follows the work

Phase 1 · 20% Phase 2 · 25% Phase 3 · 55%

20%

Phase 1 · AI-Assisted Development

Aider self-study + setup gate
Lab 01 checkoff
Project 1
Participation

25%

Phase 2 · Scripted Agentic Workflows

Labs 02–04 checkoffs
Project 2
Participation

55%

Phase 3 · Orchestrator Capstone

Labs 05–12 checkoffs
Project 3 (largest single grade)
Final demo
Participation

Phase weights are near-final. The week 1 syllabus locks them, along with the within-phase breakdown.

Grading scale

A+97–100

A93–96

A−90–92

B+87–89

B83–86

B−80–82

C+77–79

C73–76

C−70–72

D60–69

Fbelow 60

These thresholds are guaranteed minimums. Boundaries may move in your favor, never against you.

06 Policies

House rules

Academic integrity

AI use is required in this course; that is the point. Understanding is what gets graded, not generation.

You may use AI tools for all coding assignments
You must understand every line you submit and be able to explain it
You must document your AI usage in each submission: tools, models, prompts
You may not copy another student's code, prompts, or reports
You may not submit prior-semester or external work without attribution
Lab checkoffs and the final demo are where we verify understanding, in person

Violations go through the College of Engineering Honor Code process.

Attendance

Lectures: mandatory. They are built around live demos and discussion you can't get from slides.
Labs: mandatory. Checkoffs happen in person with a TA. A missed lab without prior approval scores zero for that checkoff.

Late work

Labs: checked off during the session; no late submissions
Projects: 10% penalty per day, up to 3 days; no credit after that
Extensions: for documented emergencies, case by case. Ask before the deadline, not after.

Collaboration

All three projects are individual work.
Discussing concepts and debugging together is fine in every phase. Sharing code, prompts, or reports is not.

Accommodations

Students with disabilities should contact Services for Students with Disabilities and share the accommodation letter with the instructors within the first two weeks of term.

Mental health and wellbeing

If you or someone you know is struggling, help is available.

Counseling and Psychological Services (CAPS): 734-764-8312
Dean of Students Office: 734-764-7420
Crisis Text Line: text HELLO to 741741

07 Tools + costs

What you need, what it costs

Required setup

A laptop running macOS, Linux, or Windows with WSL. No GPU needed.
Python 3.11+, git, and a GitHub account
Aider (phase 1) and Claude Code (week 3 onward)
LLM API access; provider and key setup guidance lands in week 1

Model policy

In phase 1, Aider is pinned to older, less capable models on purpose. Weaker models force the deliberate context and prompt work the phase is teaching.

Phases 2 and 3 open up tools and models. Use, understand, document, excel.

Budget

There is no textbook. Instead, plan on $100–200 in LLM API costs across the semester. Phases 1 and 2 together run about the cost of a pizza. Phase 3 can reach the price of an engineering textbook, depending on the tasks you attempt, your model selection, and how well you use rate limits and budget controls (all of which we teach).

08 Resources

Repos and communication

Course repositories

Org: github.com/eecs498-aase
Course site: eecs498-aase.github.io
Phase 1: ai-assisted-development
Phase 2: scripted-agent-orchestration
Phase 3: modular-agent-orchestrator

Communication

Discussion forum: TBD, announced in week 1
Office hours: TBD, announced in week 1
Staff email: eecs-aase-staff@umich.edu

Acknowledgments

This course draws on ideas and materials from people we're grateful to: Dan Disler (IndyDevDan), whose Principled AI Coding series shaped phase 1's pedagogy; Paul Gauthier, creator of Aider; Andrew DeOrio, for early course concept work; and Suraj Rampure, for course design feedback.