Visual Asset Management System (VAMS) is an AWS-native, open-source solution designed specifically for managing and distributing visual assets used in Physical AI and Spatial Computing applications. The system addresses a critical organizational challenge: spatial assets like 3D models, point clouds, CAD files, and geospatial data are typically large, fragmented across multiple systems, difficult to version, and inaccessible to non-technical teams. VAMS provides a unified platform that democratizes access to spatial data while maintaining data sovereignty within an organization's AWS account.

The platform supports storage and management of any file type, with native built-in support for 3D meshes in formats including glTF, OBJ, STL, and FBX, CAD models such as STEP and BREP, point clouds in E57, LAS, and LAZ formats, USD scenes, gaussian splats, documents, images, video, and audio files. The system is extensible through custom viewer plugins and processing pipelines, allowing organizations to adapt the platform to their specific requirements without vendor lock-in. Several independent software vendors have built commercial products on top of VAMS, and enterprise customers across defense, energy, manufacturing, and construction sectors have adopted and contributed to the solution.

VAMS deploys as a serverless architecture using AWS CDK with over ten nested CloudFormation stacks, leveraging core AWS services including API Gateway, Lambda, DynamoDB, S3, OpenSearch, and Cognito. The solution supports deployment to both AWS commercial regions and AWS GovCloud, making it suitable for government and defense applications. Users interact with VAMS through three primary interfaces: a web interface for interactive use and visualization, a command-line tool called vamscli for automation and scripting, and a REST API for custom integrations and programmatic control.

Key capabilities include centralized storage of 3D models and point clouds in Amazon S3 with versioning and access control, interactive browser-based visualization with seventeen built-in viewer plugins powered by technologies like Three.js, Potree, and Cesium, automated asset transformation through configurable pipelines backed by Lambda, SQS, or EventBridge, intelligent search combining full-text and metadata search powered by OpenSearch with map-based geographic views, and fine-grained permissions using attribute-based and role-based access control at both API and data entity levels.

The repository shows moderate but sustained activity, with scheurik as the most active contributor tracked by GitGenius with sixteen events, while documentation has been the most active issue label. The median response latency for issues and pull requests across tracked items is approximately 18,104 hours, indicating the project operates with extended response timelines typical of AWS Labs initiatives. The repository overlaps with the electron/electron project through shared contributors, suggesting cross-pollination of development expertise.

Use cases span multiple industries: defense and public sector applications including digital twins and training simulations with GovCloud support, energy and utilities for facility scanning and maintenance data management at terabyte scale, construction and architecture for comparing LiDAR scans against design plans and BIM management, manufacturing for digital twins and CAD centralization, robotics and Physical AI for simulation environments and training data management, digital twins with versioned facility scans and automated viewer generation, AR/VR/XR and media applications with optimized 3D content distribution and format conversion, and AI and machine learning for managing and curating training datasets with metadata tagging and automated labeling pipelines.

The solution is categorized as near-production-grade at its default configuration and is designated as an AWS Spatial Data Plane solution. Comprehensive documentation is available through a Docusaurus-based site covering deployment, user guides, CLI reference, pipeline development, API reference, and troubleshooting. The project requires Python 3.12 or higher, Docker or compatible container runtime, Node.js 20 or higher, npm, AWS CLI, and AWS CDK CLI for deployment.