Cannot sign in
Check that you are using the assigned `koicatiu.onmicrosoft.com` account. Keep working in mock mode and ask your tutor to escalate.
Azure practice environment
Azure Lab Guide
A practical guide for using your assigned Azure and Microsoft Foundry practice account during the AI-901 pathway. Keep your scope small, save evidence, and protect credentials.
Lab Workflow
1
Sign in
Use the account provided by the operations team and complete MFA when required.
2
Open Foundry
Go to Microsoft Foundry and select your own assigned project.
3
Test a model call
Run a short prompt in the playground or a minimal Python example.
4
Build the lab evidence
Connect the model to a small notebook, client, or controlled experiment.
5
Save evidence
Keep the required output, screenshot, or transcript for your AI-901 coursework.
6
Close safely
Stop idle work and ask your tutor before changing resources.
Account Setup
Use assigned account
Use the personal KOICA-TIU Azure account given by operations for your practice work.
Complete MFA
Set up the required authentication method during first sign-in. Ask a tutor if the prompt blocks your progress.
Check personal scope
Confirm that the resource group, Foundry project, and deployment belong to your account.
Keep credentials local
Put endpoint, key, and deployment names in `.env` only. Commit `.env.example`, not `.env`.
Foundry Quickstart
1
Open the assigned project
Select your personal project and confirm that your tutor can see the same account scope.
2
Find model deployment
Write down the deployment name in your local `.env`; do not put it in public screenshots if it reveals internal setup.
3
Run one small prompt
Use one short test question first. Avoid repeated long prompts while designing the UI.
4
Connect code
Use the sample template or a short Python client to call the model from your prototype.
5
Prepare fallback
Save one screenshot or short video after your first successful model response.
Usage Rules
These rules keep individual lab work attributable, reviewable, and safe for the verified credit pool.
Use your personal scope only
Work only in the individual Azure account, resource group, and Foundry project assigned to you.
Keep prompts small
Test with short prompts first. Long repeated prompts consume quota quickly.
Never expose secrets
Do not paste API keys, endpoints, passwords, or tokens into GitHub, slides, screenshots, or chats.
Keep the scope small
Stop idle work and remove temporary resources when your tutor confirms it is safe.
Ask tutors early
Report quota, login, or deployment problems before the final demo block.
Troubleshooting During a Lab
MFA prompt blocks progress
Do not retry many times. Show the screen to your tutor so the operations team can reset or verify the method.
Foundry project missing
If your personal project does not appear, stop and ask a tutor to check the account assignment row.
Model call fails
Check endpoint, deployment name, API version, and `.env` loading. Never paste keys into chat or screenshots.
Quota warning
Stop repeated tests, switch to shorter prompts, and ask the tutor to review loops or background calls.
Deployment fails
Keep the coursework evidence and backup screenshot ready while the tutor checks your assigned scope.
Student Account FAQ
Which account should I use?
Use only the KOICA-TIU Azure account assigned by the operations team.
Can teammates share one password?
No. Each student account is individual. Do not share passwords or MFA methods.
Where do I store Azure values?
Store endpoint, deployment name, and keys in `.env`. Commit `.env.example` only.
What happens after the semester?
Operations preserves academic evidence and cost records before reviewed resource reclamation.
Quota Basics
Personal allowance
Your soft allowance, spent amount, and as-of time appear in the Academic Console. Ask an administrator before requesting more.
Model usage
Use low-volume tests during development and reserve longer prompts for the final demo.
Throughput
Use the approved deployment and small calls. Do not create extra paid services without administrator approval.
Data
Use small public or synthetic data. Do not upload private personal data.
Credit Safety
Protecting the Shared Credit
KOICA-TIU operates a verified Azure pool for individual AI-901 practice. Public pages never expose live balances; use the Academic Console for your own allowance and usage.
Verified pool
Operations verified
Public pages do not display the live credit balance or any student balance.
Student allowance
Console only
Each student sees only their own allocated, spent, remaining, and as-of values.
Evidence first
Before reclamation
Academic and cost evidence is preserved before reviewed resource deallocation.
Mock-first
Use mock mode or local stubs while building the UI. Only switch to live Azure calls when you have a working prototype.
Short prompts
Iterate with short test prompts (under 200 tokens) until the flow works. Save longer prompts for the final demo.
Small samples
Use 3-5 sample documents for RAG until the retrieval works. Scale up only after evaluation passes.
Stop idle work
When a lab is complete, stop unnecessary calls and temporary resources after tutor review.
No background loops
Never leave a polling loop, recursive agent, or auto-retry running. One missed loop can burn a team budget in minutes.
Watch your allowance
The console shows your own allocated, spent, remaining, and as-of values. Alerts do not automatically suspend your account.
If you see a quota warning, do not retry repeatedly. Stop the call, save the message, and ask your tutor to review the scope and usage.
Estimate Before You Build
Plug in how many tokens you expect to use for input and output. The simulator returns the KRW cost using the verified gpt-4o-mini planning price.
Cost Simulator
Input price
$0.000165/1K tokens
Output price
$0.00066/1K tokens
Estimated cost
-
-
Default rate 1400 KRW/USD. Adjust either input box to estimate per-call or per-event cost. Hackathon teams typically run 5K-50K tokens per build, 100K-400K tokens for a full day.
Before You Present
- Your work runs in your assigned personal scope.
- Your demo does not show secrets.
- Your app has a backup screenshot or recording.
- Your tutor has checked your final demo path.