Street-level travel time estimation along with the temporal variations in patterns of travel times is an important component of traffic planning and operation in modern urban settings. In this work, we propose a scalable distributed-computing based methodology to leverage coarse-grained and aggregated travel time data to estimate the street-level travel times of a given metropolitan area. Our approach, termed TranSEC (Transportation State Estimation Capability), leverages easy-to-obtain, aggregated data sets with broad spatial coverage, such as the data published by Uber Movement and can handle road networks of very large size such as whole metropolitan areas. TranSEC is flexible enough to accommodate augmentation with fine-grained but potentially more expensive datasets, such as curated GPS-based data and probe data. Our proposed methodology uses a graph representation of the road network and combines several techniques such as weighted shortest-path routing, a trip sampling and a biased travel time sampling schemes, graph sparsification through betweenness centrality, and an iterative optimization flowthat solves successive constrained least-squares optimization problems. TranSEC further uses graph partitioning tools to enable distributed solution to the problem. We demonstrate our method on the full Los Angeles metropolitan-area where aggregated travel time data is available for trips between traffic analysis zones using a 1280-core supercomputer and visualize the temporal traffic trends at the street-level.