BC Card enhances financial AI operations system with Red Hat OpenShift AI

BC Card built a platform based on Red Hat OpenShift AI that enables the deployment and operation of AI models in an on-premises GPU environment. By using vLLM, FP8 dynamic quantization, and the SLM swarm architecture, it reduced GPU inference costs by up to 70% and increased inference speed by 2 to 3 times. The company has also achieved an annual efficiency gain of 102,000 hours in areas such as franchise screening, RPA, and call center automation.