Free Google Professional-Machine-Learning-Engineer Actual Exam Questions - Question 10 Discussion
containers, and a service that supports autoscaling and monitoring for online prediction requests.
Which platform components should you choose for this system?
The question highlights autoscaling and monitoring for online predictions, so B fits best.
B/D? B nails scheduled retraining and deployment with autoscaling, but D covers custom containers for training too. If Docker’s needed just for deployment, B fits better; otherwise, D could be the safer bet.
B seems right since Vertex AI Pipelines handle retraining and AI Platform Prediction supports deployment with autoscaling.
Makes sense to go with B here. Vertex AI Pipelines handle the scheduled retraining part smoothly, and AI Platform Prediction supports Docker containers for deploying models with autoscaling and monitoring. The question doesn’t specify custom container training, so using AI Platform Prediction covers both container support and serving needs well. D includes Cloud Composer, which is more for workflow orchestration but doesn't directly support autoscaling or monitoring of prediction requests like AI Platform Prediction does, so B seems like the better fit overall.
Good point about the autoscaling and monitoring specifics. I think option B fits best because AI Platform Prediction is designed for exactly those online prediction needs, plus Vertex AI Pipelines handles scheduled retraining smoothly. B
Could App Engine really handle autoscaling and monitoring better than AI Platform Prediction here?
Option B fits best since AI Platform Prediction handles autoscaling and monitoring well.
B/C? I get why B is popular since AI Platform Prediction handles autoscaling and monitoring, plus Vertex AI Pipelines is good for scheduling retrains. But C also has BigQuery ML, which might simplify some model training with SQL and scheduled pipelines via Cloud Composer. The downside: BigQuery ML might not support custom Docker containers like they want. So it depends if the priority is container support or ease of scheduled retraining with monitoring. D seems overcomplicated and A’s App Engine might not fully cover autoscaling for predictions.
B imo since Vertex AI Pipelines schedule retraining well and AI Platform Prediction manages autoscaling.
It’s A because App Engine can autoscale and manage online prediction requests, while Vertex AI Pipelines handle the scheduled retraining. Plus, App Engine supports Docker containers naturally, covering all bases here.
This one’s tricky, but I’d go with B. Pipelines cover scheduled retraining nicely, and AI Platform Prediction supports Docker containers while handling autoscaling and monitoring smoothly. App Engine in A isn’t really built for model serving at scale like AI Platform Prediction is, so that rules out A and D. C uses BigQuery ML which isn’t really about custom containers or autoscaling online predictions, so that’s out too. So B fits the requirements best without extra overhead.
B/D? AI Platform Prediction is the clear choice for autoscaling and monitoring, but if custom containers are needed for training, D could work too. Depends how much container support is needed beyond prediction.
Maybe B makes the most sense since AI Platform Prediction handles autoscaling and monitoring well, plus it supports Docker containers for serving models. Pipelines can schedule retraining easily too.
Option B fits since AI Platform Prediction handles autoscaling and monitoring best.
B. App Engine doesn't support custom containers as well as AI Platform Prediction.
Maybe B works since Vertex AI Pipelines handles scheduled retraining, and AI Platform Prediction supports Docker containers for online predictions with autoscaling and monitoring. App Engine might not be needed here.
D imo since Cloud Composer handles scheduling, custom containers fit training, and App Engine autoscale well.
B/C? The question asks for scheduled retraining, Docker containers, plus autoscaling and monitoring for online predictions. Vertex AI Pipelines definitely covers scheduled retraining well. Between B and C, BigQuery ML feels more focused on training and querying models directly in SQL, but it might lack the full autoscaling prediction service mentioned. AI Platform Prediction in B offers built-in autoscaling and monitoring for predictions, which fits better with the requirements. So I’d rule out C because it seems less complete for the online prediction part. B looks like a tighter fit overall.
Anyone know if the prediction service needs to be serverless or can it run on managed VMs? That might change the best option here.