This artifact provides a reproducible experimental framework for evaluating TCP congestion control algorithms in public high-performance wide-area networks (HP-WANs), targeting large-scale data transfer scenarios such as scientific workflows, distributed AI training, and GPU-as-a-service, where predictable performance and efficient bandwidth utilization are critical. It constructs a multi-site topology across four geographically distributed FABRIC sites interconnected via bandwidth-reserved L2PTP tunnels (20 Gbps by default), with sender and receiver nodes and routers running a custom DPDK-based traffic shaper (tm10) to enable controlled forwarding and shaping across multiple virtual circuits. The entire setup is automated through a Jupyter notebook that performs dynamic resource discovery, slice reservation, kernel configuration, DPDK setup, VLAN-based tunnel multiplexing, and end-to-end routing configuration, allowing flexible experimentation across different congestion control algorithms, flow configurations, and network conditions. The framework is used to evaluate TCP CUBIC, BBRv1, and BBRv3. Performance is evaluated using two key metrics: FCT efficiency, which measures how closely observed flow completion times match the ideal transfer time, where higher efficiency indicates more predictable transfers, and bandwidth overhead, which quantifies retransmission-induced inefficiency. The artifact includes all scripts, configurations, and example outputs required to reproduce the experiments and generate plots.