Skip to content

Defining Native Applications

A Native Application is described by an application profile — a YAML document with applicationProfile at its root — that defines the workload (container, VM, or binary), its runtime environment, and the requirements that govern where and how it runs.

Profile Structure

Every application profile follows this layout:

applicationProfile:
  metadata:
    # application identity and annotations
  specs:
    # runtime, resources, network, constraints, qos
  status:
    # runtime state, updated dynamically
Field Type Required Description
applicationProfile.metadata object ✅ Application metadata — see metadata
applicationProfile.specs object ✅ Application requirements — see the specs sections below
applicationProfile.status object ❌ Runtime state, updated dynamically — see status

Field Reference

Below we provide all mandatory (✅) and optional (❌) fields that are registered in the Hyper-AI platform.

metadata — Application Metadata

Field Type Required Example Description
type string, native ✅ "native" Application profile type: integration, native, or device.
schemaVersion string ✅ "1.1.0" latest App profile schema version using semantic versioning (e.g., 1.0.0).
name string ✅ "hello-world-webserver" Globally unique application name for identification.
version string ✅ "1.0.0" Internal version identifier for the application instance.
description string ❌ "Nginx web server for HyperAI platform assessment." A short description of what the application does.
owner string ✅ "UoA-Team" Person or organization responsible for this application.
lifecyclePhase string, enum: development testing production ✅ "development" Lifecycle stage: development, testing, or production.
createdAt string (RFC 3339) ✅ "2026-06-04T12:00:00Z" Creation timestamp in RFC 3339 format (UTC).
updatedAt string (RFC 3339) ✅ "2026-06-04T12:30:00Z" Last modification timestamp in RFC 3339 format (UTC).
annotations.intent string ❌ "platform-demo" Purpose of the application
annotations.domain string ❌ "education" Classification of application (e.g., AI, Automotive, robotics)
annotations.language string ❌ "Python 3.10" Programming language used by the application
annotations.dependencies list ❌ ["fastapi", "uvicorn"] Required software libraries (e.g., ML frameworks)

specs.runtime — Runtime Requirements

Field Type Required Example Description
executionType string, enum: container vm ✅ "container" Type of execution platform (e.g., container, vm, native, android)
entryPoint string ✅ "uvicorn" Path to the main script or command executed at container startup
args list ✅ ["main:app", "--port", "8000"] Arguments passed to the entry point command
baseOS.name string ✅ "python" Name of base OS (e.g., ubuntu)
baseOS.version string ✅ "3.10-slim" Version of the base OS (e.g., 20.04)
containerImage.uri string ✅ "registry.example.com/org/app" OCI-compliant container image URI
containerImage.tag string ✅ "latest" Optional image tag (containers)
hypervisor string, enum: qemu kvm xen vmware ❌ "kvm" Hypervisor/manager (qemu, kvm, xen, vmware)
acceleratorRuntime[].name string ❌ "cuda" Name of the runtime environment for accelerators
acceleratorRuntime[].version string ❌ "12.2" Version of the accelerator runtime

specs.resources — Resource Requirements

Field Type Required Example Description
cpu string (>0, millicores) ✅ "2000m" Number of CPU cores required (e.g., '1000m' for 1 core).
memory string (Mi Gi Ti) ✅ "10Gi" Memory requirement.
storage string (Mi Gi Ti) ✅ "1Gi" Storage requirement (if needed).
gpu integer (≥0) ❌ 1 Required GPUs.
tpu integer (≥0) ❌ 0 Required TPUs.
accelerators integer (≥0) ❌ 0 List of FPGAs, ASICs

specs.network — Network Requirements

Field Type Required Example Description
ports[].port integer ✅ 8000 port number
ports[].protocol string ✅ "TCP" protocol used by this port
ports[].publicExposure boolean ❌ true Whether the port is publicly exposed
networkBandwidthMin string (Mbps Gbps) ❌ "100Mbps" Minimum network bandwidth.

specs.constraints — Scheduling Constraints

Field Type Required Example Description
supportedArchitectures list ✅ ["x86_64"] Compatible CPU/GPU architectures.
trustScore string (1–5) ❌ "4.5" Reliability and security score.
geoLocationRequirement string ❌ "Europe-only" Location restrictions.
isHighlyAvailable boolean ❌ false Specifies if deployment requires redundancy.
faultTolerance string ❌ "restart" Resiliency strategy (e.g., restart policy).
securityLevel string (1–3) ❌ "high" Security classification (e.g., 3: high, 2: medium, 1: low).
dataClassification string ❌ "private" Data sensitivity

specs.qos — Quality of Service

Field Type Required Example Description
latencyToleranceMax string (ms) ❌ "150ms" Maximum tolerable latency.
energyCost string (kWh) ❌ "0.5kWh" Estimated energy consumption per hour.
monetaryCost string (currency) ❌ "0.05" Deployment cost per hour.
resilience string ❌ "restart" Tolerance and failure mechanism (e.g., restart)
availability string (%) ❌ "99.0%" Uptime SLA target
startupTime string (secs) ❌ "10s" Time to become operational

Note: resilience: Redundant to faultTolerance, will be removed in the next schema version.

status — Runtime Status (Dynamic)

Unlike the fields above, status is not set in the profile — it reflects the runtime state of the application and is updated dynamically.

Field Type Required Example Description
isOnline boolean ✅ true Whether the application is currently active
lastHeartBeat string (RFC 3339) ✅ "2026-06-04T12:30:00Z" Last time the app sent a status signal
runtimeState string ✅ "running" Current runtime status of app (e.g., running, failed)
isStateless boolean ❌ false Indicates if the application maintains state.
resourceUsage.cpu string ✅ "2000m" CPU usage (e.g., 2000m)
resourceUsage.memory string ✅ "512Mi" RAM usage
resourceUsage.storage string ✅ "1Gi" Storage consumed
resourceUsage.gpu integer (≥0) ❌ 0 Number of GPUs consumed
resourceUsage.tpu integer (≥0) ❌ 0 Number of TPUs consumed
resourceUsage.accelerators integer (≥0) ❌ 0 Number of FPGAs, ASICs used

Examples

See Cookbook & Examples for minimal native app examples.