AI systems foundations · Lesson 2 of 15 · 6 min

Separate model and application

Model weights produce predictions; the application owns routing, authentication and presentation. A change to one layer does not automatically change the other.

Open the lab Course map

Mastery 0/3Run itExplain itApply it

learner@clairos:~ $

↑ history · Tab complete · Ctrl+C cancel · Esc, Tab to leaveExit 0

Your objectives

Hints are always available. Wrong turns never cost XP.

1
Read the case evidence
Read /opt/ai-lab/cases/aiops-systems-foundations-002.json. Separate the recorded facts from your assumptions.
Hint & worked example
Model weights produce predictions; the application owns routing, authentication and presentation. A change to one layer does not automatically change the other.
The complete case JSON is read successfully.
```
cat /opt/ai-lab/cases/aiops-systems-foundations-002.json
```
Common mistake to avoid: Re-download m3 because every application regression is a weights problem.
2
Isolate the deciding signal
Select the changed_layer field from the case JSON. What does it establish—and what does it not establish?
Hint & worked example
Review gateway-12 because the application changed while the weights stayed at m3.
A successful JSON query isolates changed_layer.
```
jq .changed_layer /opt/ai-lab/cases/aiops-systems-foundations-002.json
```
Common mistake to avoid: Using a different case or an unrelated signal as the justification.
3
Record your operational decision
Choose the justified code: review-application, change-without-evidence, or ignore-the-signal. Save that one code followed by a newline in /home/learner/lab/aiops-systems-foundations-002-decision.txt.
Hint & worked example
Review gateway-12 because the application changed while the weights stayed at m3.
The decision artifact contains the correct code, not just a matching command in history.
```
printf '%s\n' 'review-application' > /home/learner/lab/aiops-systems-foundations-002-decision.txt
```
Common mistake to avoid: Re-download m3 because every application regression is a weights problem.

Understand the pattern

Model weights produce predictions; the application owns routing, authentication and presentation. A change to one layer does not automatically change the other.

Interpret the evidence

Review gateway-12 because the application changed while the weights stayed at m3.

A tempting mistake

Re-download m3 because every application regression is a weights problem.

All names, measurements, prices, versions, and thresholds in this exercise are fictional training data. The evidence supports this bounded scenario, not a claim about a live provider.

Worked example and expected result

$ cat /opt/ai-lab/cases/aiops-systems-foundations-002.json

{
  "fixture": "Fictional training data; not a live measurement.",
  "weights_revision": "m3",
  "app_revision": "gateway-12",
  "changed_layer": "application"
}

Exit status: 0. The command completed successfully in the training fixture.

$ jq .changed_layer /opt/ai-lab/cases/aiops-systems-foundations-002.json

"application"

Exit status: 0. The command completed successfully in the training fixture.

$ printf '%s\n' 'review-application' > /home/learner/lab/aiops-systems-foundations-002-decision.txt

(no output; exit 0)

Exit status: 0. The command completed successfully in the training fixture.

Try a variation

Name the assumption that would have to change for “review-application” to stop being the justified action. Do not silently replace missing evidence with a guess.

Explain it

What does the evidence in “Separate model and application” justify?Make the change now and gather the missing evidence afterward.Review gateway-12 because the application changed while the weights stayed at m3.Re-download m3 because every application regression is a weights problem.

Apply it elsewhere

A colleague concludes: “Re-download m3 because every application regression is a weights problem.” What should you do?Reject that conclusion; return to the case evidence and its stated limits.Accept it because the training command completed.Treat this fixture as a live measurement.

Keep the momentum.

Complete the lab, then explain and apply the pattern.

Next: Trace a request lifecycle Try an incident drill

Anonymous lesson and mastery progress lasts for this session. Sign in to keep synced lesson XP between sessions; mastery stays device-local. No speed bonus, lost lives, or hint penalties.

Keep learning

Case methodology: evidence, limits, decision