Success story

End-to-end media asset management platform

AI-assisted media discovery, hybrid scaling, and zero vendor lock-in

Quick project overview

Client
European broadcast software provider
Business challenge
Build a next-gen MAM platform with AI-assisted metadata enrichment, hybrid infrastructure, and configurable media workflows
Oxagile’s involvement
Oxagile owned the engineering of about 10 microservices, including media processing and infrastructure

Platform performance targets

Under 500 msFast, secure media search
Under 60 secondsRapid file ingest completion
Up to 20 bln valuesEnterprise-scale metadata capacity

The client’s vision for the MAM solution

End-to-end media ingest and validation
We need a single MAM platform to ingest, validate, and manage media files from external capture systems and production teams, including recordings that are still in progress.
AI-powered content enrichment and search
Editors need to tag events, create clips, and find relevant footage using searchable metadata, including AI-generated information supplied by external services.
Content review and automated distribution
Reviewers need to inspect media content, record compliance decisions, and trigger follow-up workflows. Approved files can then be routed to editing systems, shared storage, and other production environments through configurable workflows.

Inside the MAM solution

Manage

Core asset management

Centralize media assets, versions, renditions, clips, and collections
Enrich assets with flexible metadata and structured content hierarchies
Control access through role- and attribute-based permissions

Process

Media ingest and processing

Ingest media through watch folders and manual uploads, with support for growing files from external capture systems
Automatically validate files, generate proxies, and assemble assets
Orchestrate transcoding, quality control, archiving, and delivery through no-code workflows
Manage storage, replication, retention, and file lifecycle policies

AI-assisted media discovery

Integrate metadata generated by external AI services into the MAM platform
Make AI-generated metadata available for asset discovery and filtering
Find relevant footage using keyword search, metadata filters, and faceted navigation
Enforce access permissions across all search results

Automate

Integration and workflow automation

Connect newsroom systems, editing tools, QC platforms, and production storage through supported integrations
Automate media processing through configurable workflows and integrations
Trigger notifications for processing events, validation failures, and approval requests
Coordinate automated and human-driven tasks across the media lifecycle

Operate

User experience and operational control

Provide a unified, role-based interface for editors, loggers, administrators, and distribution teams
Use frame-accurate proxy playback, clip marking, and timeline navigation
Customize workspaces, saved searches, and self-service dashboards
Manage tenant configuration, monitor system health, and troubleshoot operational issues

Core MAM modules we built

Oxagile delivered a substantial part of the MAM platform’s media processing and cross-cutting infrastructure, owning nine backend modules and a set of dedicated frontend micro-frontends.

Backend modules

Watch-folder and manual file ingest, growing-file support, and automated file stability checks
Storage abstraction, intelligent access routing, storage-tier transitions, and automated imports
No-code workflow designer and durable execution engine for media process orchestration
Integration with external AI enrichment services to incorporate AI-generated metadata into MAM workflows
Integration components for connecting MAM workflows with external production systems
Event-driven triggers and multi-channel notification delivery
Centralized global and tenant-specific configuration, with DevOps support for tenant provisioning
Centralized metrics, logs, and distributed tracing for monitoring and troubleshooting
Configurable, self-service operational dashboards

Frontend engineering and architecture

Dedicated micro-frontends for workflow orchestration, ingest, storage management, proxy playback, and dashboards
Reusable CRUD interface patterns for AI Services, Integrations, Notifications, and Configuration
Modular frontend architecture within the shared MAM application
Collaborating on BFF API contract and shared architectural boundaries
Coordination with the Search team on index schemas and query interfaces for compatibility with Catalog and Metadata services

The team behind the scope

Project Manager
Business Analysts
Backend Developers
Frontend Developers
Manual QA Engineer
Automation QA Engineers
DevOps Engineers
UI/UX Designer
Architect

Engineering challenges and architectural decisions

Engineering challenges and architectural decisions
Area	Challenge	Architectural decision
Media processing at scale	Reliable ingest of high-volume media, including growing files that must be available before recording finishes.	Watch-folder monitoring and manual upload mechanisms, with asynchronous processing and early proxy generation to make growing files available before recording completes.
Resilient workflow orchestration	Long-running media workflows spanning QC, transcoding, AI enrichment, archiving, and delivery, with failures and human approvals.	A no-code workflow designer backed by Temporal for durable execution, retries, recovery, and coordination of long-running tasks.
Heterogeneous system integration	Connecting MAM workflows with external production and storage systems that use different protocols, APIs, and data formats.	Capability-based adapters, reliable event-driven exchange, and secure, compliant external integrations.
Modularity and operational visibility	Enabling independent module development and maintaining a coherent user experience and visibility across distributed services.	Domain-oriented backend services, independently deployable micro-frontends, reusable UI patterns, and vendor-neutral observability through OpenTelemetry.

Security best practices applied

Authentication and identity
Keycloak-based OIDC with PKCE, mandatory MFA, secure token rotation, and enterprise identity federation.
Zero-trust authorization
Combined RBAC/ABAC, pre-query entitlement filtering, and server-side field-level access control across Catalog, Search, and AI retrieval.
Multi-tenant isolation
JWT-verified tenant identity, database-per-tenant-per-microservice isolation, and tenant-scoped Kafka and Temporal operations.
Data protection and secrets management
TLS/mTLS, encryption at rest, vault-managed credentials, file integrity checksums, and tenant-specific secret rotation.
Privacy and auditability
Mandatory PII redaction in logs and traces, metric cardinality safeguards, and immutable audit trails for every mutation.
Secure integrations and frontend
SSRF-resistant outbound gateways, signed and rate-limited webhooks, hardened sessions, CSP, trusted micro-frontends, and tenant-scoped WebSockets.
Secure supply chain and delivery
SBOM, SAST/DAST, signed container images, protected deployment credentials, and contract-tested, backward-compatible migrations.
Reliability as security
Transactional outbox, idempotent processing, replay protection, circuit breakers, and bounded retries to prevent duplicate actions and data loss.

Tech stack

Backend: .NET 8/10+ · C# · PostgreSQL · Elasticsearch · Apache Kafka · Temporal · Keycloak
Frontend: React 19 · TypeScript · Nx · Module Federation 2.0 · Vite · Material UI 7 · TanStack Query
Infrastructure and observability: Kubernetes · OpenTelemetry · GitOps-based CI/CD

More MAM projects

MAM orchestration solution
Optimize and automate complex workflows
MAM integration and migration
Complex workflows and system knowledge gaps tackled
Turning MAM system complexity into clarity
Uninterrupted MAM services under critical workloads
Developing a media review platform
Video engineering expertise, advanced MAM workflows, and secure playback