Lab “Monitoring Vertex AI Models” – Step 1 dependency conflicts (BigQuery 1.x vs AI Platform SDK 1.51.0)

[anhxuanpham]

# 🐞 Bug Report: Lab “Monitoring Vertex AI Models” – Step 1 dependency conflicts

## 📚 Context
While running the lab [Monitoring Vertex AI Models](https://www.cloudskillsboost.google/course_templates/9/labs/588147), the initial setup in **Step 1** fails due to incompatible dependencies.

---

## 📂 Affected Notebook
[`courses/machine_learning/deepdive2/production_ml/labs/model_monitoring_for_custom_model_batch_prediction_job.ipynb`](https://github.com/GoogleCloudPlatform/training-data-analyst/blob/master/courses/machine_learning/deepdive2/production_ml/labs/model_monitoring_for_custom_model_batch_prediction_job.ipynb)

---

## ⚠️ Problematic Install Command (from lab instructions)
```bash
!pip install --use-deprecated=legacy-resolver \
  google-cloud-bigquery==1.27.2 \
  pandas==1.5.3 \
  pandas-gbq==0.17.9 \
  pyarrow==10.0.0 \
  tensorflow-data-validation==1.14.0 \
  google-cloud-aiplatform==1.51.0 \
  tensorflow-metadata==1.14.0 \
  google-api-core==1.34.1 \
  google-auth==2.26.1 \
  google-cloud-core==1.7.3 \
  google-resumable-media==1.3.3 \
  tfx-bsl==1.14.0

---

❌ Issue

• google-cloud-bigquery==1.27.2 requires google-auth <2.0
• google-cloud-aiplatform==1.51.0 requires google-auth >=2.14.1

This creates a hard conflict.
Even if installed with --use-deprecated=legacy-resolver, imports fail with errors like:

TypeError: Descriptors cannot be created directly.
If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc >= 3.19.0.

---

🔁 Steps to Reproduce

1. Start the lab “Monitoring Vertex AI Models.”
2. Run the Step 1 pip install command from the instructions.
3. Try to import google.cloud.aiplatform.
4. Observe the crash.

---

✅ Expected Behavior

Dependencies should install cleanly so learners can proceed with the lab.

---

🐞 Actual Behavior

Conflicting requirements for google-auth prevent the lab from running at all.
This blocks progress right at the beginning of the exercise.

---

💡 Suggested Fix

• Update dependencies to be internally consistent. Options:

  • Upgrade BigQuery client to 3.x (compatible with google-auth >=2.14.1).
  • Or downgrade google-cloud-aiplatform to a version compatible with google-auth <2.0.

• Provide a tested requirements.txt or prebuilt environment for learners.

---

📌 Impact

This prevents learners from completing the lab and blocks progress in the Cloud Skills Boost / Quan Quan GCP track.

---

🏷️ Labels

• bug
• dependencies
• lab

---

<img width="2560" height="1440" alt="Image" src="https://github.com/user-attachments/assets/6eb5a467-8ebc-43a2-8005-94bf72c240a6" />