About this project

VModal AI for Android is a UI-free Kotlin SDK that connects Android applications to the VModal video and image search service. Apps can configure a client with a project ID and runtime API key, scope requests to collections and streams, upload videos, create or check image indexes, and search for visual moments by semantic text or imagery. Search results can be resolved in bulk for display in the application's own interface. The SDK handles gateway communication, typed request and response models, signed streaming uploads, upload progress, cancellation, and response parsing. It exposes coroutine-based APIs and cold upload Flows, with callback-based UploadHandle compatibility mentioned for existing code. Background uploads can integrate with AndroidX WorkManager through CoroutineWorker, and user-selected files are supported as content:// URIs through the Storage Access Framework. The library is designed to be used from Jetpack Compose, classic Android Views, or custom design systems. It does not impose login screens, navigation, themes, or other UI components. Credentials are intended to be supplied at runtime and kept in memory; the documentation covers API-key management, authenticated identity calls, account-switch cleanup, typed failures, lifecycle-aware collection, and cancellation of SDK resources. Included documentation provides an integration cookbook, technical introduction, coroutine and WorkManager guidance, search-app walkthrough, and API reference. The repository also contains progressive examples: starter operations, an upload-to-search app, a complete search application, index-organization patterns, and a Framebase-style street-video search flow. The documented toolchain uses Kotlin 1.9 or newer, Java 17, Gradle 8.6, compile SDK 34, and a minimum Android API level of 24. The project is licensed under Apache 2.0.