VFIG: Vectorizing Complex Figures in SVG with Vision-Language Models | ScienceToStartup | ScienceToStartup