Google Professional Machine Learning Engineer (PMLE) 덤프 및 해설

You deployed a TensorFlow recommendation model to a Vertex AI Prediction endpoint in us-central1 with autoscaling enabled. Over the last week, you observed sustained traffic of ~1,200 requests per hour (about 20 RPS) during business hours, which is 2x higher than your original estimate, and you need to keep P95 latency under 150 ms during future surges. You want the endpoint to scale efficiently to handle this higher baseline and upcoming spikes without causing user-visible latency. What should you do?

You plan to fine-tune a video-frame classifier via transfer learning using a pre-trained ResNet-50 backbone. Your labeled dataset contains 18,000 1080p frames, and you will retrain the model once per day; each training run completes in under 60 minutes on 4 V100 GPUs, and you must minimize infrastructure cost and operational overhead. Which platform components and configuration should you choose?

Your team is preparing to train a fraud detection model using data in BigQuery that includes several fields containing PII (for example, card_number, customer_email, and phone_number). The dataset has approximately 250 million rows and every column is required as a feature. Security requires that you reduce the sensitivity of PII before training while preserving each column’s format and length so downstream SQL joins and validations continue to work. The transformation must be deterministic so the same input always maps to the same protected value, and authorized teams must be able to decrypt values for audits. How should you proceed?

Your team deployed a regression model that predicts hourly water usage for industrial chillers. Four months after launch, a vendor firmware update changed sensor sampling and units for three input features, and the live feature distributions diverged: 5 of 18 features now have a population stability index > 0.25, 27% of temperature readings fall outside the training range, and production RMSE increased from 0.62 to 1.45. How should you address the input differences in production?

You are a data scientist at a city transportation agency tasked with forecasting hourly bike-share demand per station to optimize rebalancing. Your historical trips table in BigQuery contains 24 months of data (~22 million rows) with columns: timestamp, station_id, neighborhood, weather_condition (sunny/rainy/snow), special_event (boolean), and surge_pricing_flag (boolean). You need to choose the most effective combination of a BigQuery ML model and feature engineering to minimize RMSE while capturing weekly/seasonal patterns and handling multiple categorical variables; what should you do?