Technology Development Research Project

Each student turns a problem they choose into a working service and learns to explain why it works. We build with Claude Code, in class, every week.

Course code
M04112201
Term
Fall 2026
Level
Year 3 · 3 credits
Time
Wed 15:00-17:50 (periods 7, 8, 9)
Room
Room 2605
Instructor
Donghyun Lee
Program
Social Science & AI, HUFS
  • No exams
  • Individual projects
  • Built in class every week

What you will build

You leave the course with two things. Both run at a public URL, and other people can use them.

  • A personal homepage with your introduction, interests, and links to your work. You build it in weeks 2-3 and keep adding your project and retrospective until the end of the semester.
  • A working service for a user and a problem you choose. You build the core flow, test it with real users, fix what breaks, and deploy it.

What you will learn

  • Define the user, the problem, the core flows, and a definition of done before writing code.
  • Write requests that state the goal, the context, and the completion criteria, then read and run the agent’s changes before accepting them.
  • Diagnose failures by reproducing them and checking boundary conditions and logs, and confirm each fix under the conditions that exposed it.
  • Handle data sources, personal information, and secrets responsibly, and collect feedback from at least three real users.
  • Deploy a reproducible service with a README, and explain your design decisions, your verification evidence, and the remaining limits.

What the course leaves out

  • Midterm and final exams
  • Group projects with a shared grade
  • Lectures that walk through programming syntax in order
  • Summary reports on theory
  • Paid APIs, payments, or complex authentication
  • Projects that handle real sensitive personal data

A 150-minute class

From week 5, each class starts with a lecture and a live demo of one technique, then turns to building.

Part Minutes What happens
Lecture and demo 55 The week’s technique, its limits, and the criteria for today’s work
Build together 45 Everyone builds the same example, step by step
Studio group practice 45 Your group applies the technique, led by the instructor
Personal log 5 What you did, how you checked it, what comes next

Breaks are extra. From week 7 the group practice time goes to your own project. The log is three lines, and you do not submit your full AI conversation.

Project tracks

Pick one of four directions. Each starts from a social science interest and ends in a tool people can use.

Track Examples
Public and social data explorers Youth employment and population change, living indicators for foreign residents, commuting time and cost
Campus and community information Scholarships and extracurriculars, exchange-student life, campus facilities and events
Social science research tools Survey CSV checker, interview tagging board, literature evidence cards
Comparison and simulation tools Commuting options, student event budgets, time-use scenarios

Weekly plan

The work grows in size as the semester goes on. Weeks 8 and 15 have no class, only a submission deadline.

  1. Week 1
    Course format and AI use

    What you will produce, how individual grading works, and what you are responsible for explaining.

  2. Weeks 2-3
    Personal homepage and web basics

    Information structure, HTML and CSS basics, writing and checking AI requests, Git, a README, and your first deployment.

  3. Weeks 4-6
    Services that collect and show data

    Team exercises with scheduled GitHub Actions updates, Google Sheets, and open APIs from Korea’s Public Data Portal. You also pick your own project and set its scope.

    Week 5: personal homepage due
  4. Weeks 7-8
    Core features and midterm MVP

    Build a deployed version whose core flow works, and test it with normal input, empty results, and errors.

    Week 8 (Oct 21): midterm MVP due by 11:59 AM
  5. Weeks 9-12
    Growing your service on Vercel

    Live queries with serverless functions, charts and maps, features that collect user input, optional generative AI, and a security check.

  6. Weeks 13-14
    Verification and polish

    Get peer review on preview deployments, act on feedback from users, measure performance and accessibility, and get your final submission ready.

    Week 13: beta check
  7. Week 15
    Final submission and retrospective

    Submit your final deployment, with your verification log and retrospective written up in the README.

    Week 15 (Dec 9): final project due by 11:59 AM

Grading

There are no exams, and every item is graded individually. Design taste, lines of code, commit counts, and prompt counts do not affect your grade. Every project is graded on the same criteria.

When Item What we check Weight
Ongoing Attendance University attendance rules 10%
Ongoing Participation In-class development, three-line logs, peer feedback 10%
Week 5 Personal homepage Public URL, repository, responsive layout, README 20%
Week 8 Midterm MVP A deployed build with a working core flow, and how you verified it 20%
Week 15 Final project Final deployment, repository, verification log, retrospective 40%

Student projects

Homepages and services from this course, shared with each student's permission. Students run their own sites, and any project comes down on request.

Donghyun Lee AI Group Example A sample card showing how student projects will appear here once students give permission. donghyunlee-lab.github.io/

Is this course for you?

The course fits you if you:

  • want to build a tool for a problem you have run into
  • want a public project to show before you graduate
  • want to use AI for development and still judge the results yourself
  • learn more from building and fixing than from exams

What you need:

  • Your own laptop.
  • Access to Claude Code, through a supported subscription or institutional access. We set up installation and login together in week 1.
  • No prior web development experience. Week 2 starts from how the web works, then HTML and CSS.
  • Readiness to build during class every week.

FAQ

I have never written code. Can I take this course?

Yes. In week 2 we cover how the internet works and how to edit HTML and CSS, and you build your first homepage in that class. You cannot hand in AI output as is, though. You need to run it, check it, and explain what you changed and why.

Is this a group project?

No. Deliverables and grades are individual. You sit in studio groups of three or four, try each other’s work, and give feedback, but there are no shared grades or group reports.

Can I use AI? How much?

You can use Claude Code and official documentation on every assignment. You are responsible for running and verifying the results and for explaining your design and changes. You do not submit your full chat history. Keep secret keys and real personal data out of prompts, repositories, and deployments.

I don’t know what to build yet.

You pick two candidates in week 5 and narrow them to one at the week 6 scope check. To start, you need one type of user and one problem. A scope you can finish this semester matters more than a big idea.

How much time outside class does it take?

Most development happens during the 150-minute class. Deliverable weeks (5, 8, and 15) need extra preparation.

Which tools will I learn?

In the first half you connect Google Sheets and open APIs from Korea’s Public Data Portal to pages on GitHub Pages, kept up to date by GitHub Actions. In the second half you move your project to Vercel, use serverless functions for live queries and user input, and add charts and maps. Everything stays within free tiers, and generative AI features are optional.

Based on the Fall 2026 offering. The schedule may change.

Contact: donghyun.lee@hufs.ac.kr