imgproxy is an open source, self-hosted server designed for fast and secure on the fly image processing. Its primary function is to resize, transform, and convert remote images in real time, acting as a proxy between an image source and a web application. The server processes images based on URL parameters, eliminating the need to store multiple pre generated versions of the same image. It is built on the libvips image processing library, which contributes to its low memory usage and high processing speed, capable of handling most images in under 100 milliseconds. Key features of imgproxy include support for a wide range of image formats, automatic format negotiation (such as serving WebP when supported), URL signing for security, and protection against image bombs and other vulnerabilities. The server also offers advanced capabilities like image classification, object cropping, auto quality adjustments, SVG minification, and RAW format support. It provides observability and monitoring features, and it is designed to be deployed easily via environment variables and containerized environments like Heroku or Docker. Typical use cases for imgproxy involve integrating it into content management systems, e-commerce platforms, or social media infrastructure to dynamically optimize images for different screen sizes, devices, or network conditions. The workflow generally involves configuring the server with a signing key and security settings, pointing it to remote image URLs, and then generating processed images via simple URL parameters. This approach reduces storage costs, simplifies the processing pipeline, and improves delivery performance by avoiding the creation of billions of stored thumbnails. Technical requirements for running imgproxy include a server capable of hosting the application, which can be deployed as a standalone server. The service is self-contained and does not require external dependencies for its core image processing functionality. It has been adopted by organizations with high traffic volumes, such as eBay and Medium, to securely and reliably handle billions of image transformations without significant ongoing maintenance or compute costs.