Simulation Log Collection and Result Aggregation

You must design a metrics pipeline for a large-scale autonomous-vehicle simulation platform. Every simulation run executed by the worker fleet emits a log stream containing vehicle state, perception outputs, planner decisions and collision flags. The pipeline must produce aggregated metrics keyed by scenario cluster and model version, while still allowing a drill-down into any individual run. The workload ranges from thousands up to millions of runs each day, and a single…