Component catalog

Look up the components a BullSequana AI 2.0.0 installation deploys, their role, and the page that describes each.

Agentic Friendly

Component pages explain implementation and operational relationships. Start with a product capability page unless you are deploying, operating, or troubleshooting a specific component.

This catalog lists the components of a 2.0.0 installation by product domain. Each row links to the component's own page, or to the page that describes it when it has none. Optional components are deployed only when the site enables them. For the version of each component, see Selected component updates in the release notes.

The domain indexes group the same pages by capability: AI components, Data components, and Foundation components.

AI

ComponentRoleDescribed in
PortalWeb experience for OmniChat, Studio, and AdministrationPortal
BSQAI APIAPI for chat, agents, models, files, data, usage, and administration; it also runs the document and speech workersBSQAI API
MCP GatewayAuthenticated entry point for every Model Context Protocol callMCP Gateway
MCP Lifecycle OperatorRuns Managed MCP Servers in tenant namespacesMCP Lifecycle Operator
Agent sandbox controllerRuns one isolated workspace per chat for agent code and Skill scripts. Optional.Agent Sandbox
MLflowExperiment tracking and model registry, for the Global Organization and as each tenant's private MLflowMLflow
KServeReconciles language-model and predictive-model deploymentsKServe
DoclingConverts uploaded documents for retrievalDocling
SpeachesSpeech-to-text and text-to-speech service and serving runtimesSpeaches
MilvusVector store for retrievalMilvus
OpenTelemetryInstrumentation standard of the AI servicesOpenTelemetry

Data

ComponentRoleDescribed in
Managed ingestionMoves records between source and destination connectionsManaged data ingestion
LakekeeperApache Iceberg REST catalog, one project per tenantLakekeeper
TuplesData authorization service backed by OpenFGATuples
Spark OperatorRuns Spark clusters and jobs in tenant profile namespacesSpark Operator
KubeflowNotebooks, pipelines, and profile namespaces for data scienceKubeflow
Kafka and StrimziEvent streaming, deployed and managed by the Strimzi operatorKafka
SupersetDashboards and business intelligenceSuperset
MetacontrollerDeclarative custom controllersMetacontroller

Foundation

Inference

ComponentRoleDescribed in
vLLMLanguage-model inference enginevLLM
FasterWhisperSpeech transcription engine inside SpeachesFasterWhisper
Envoy AI GatewayInternal inference gateway between the BSQAI API and model workloadsRequest path
Gateway API Inference ExtensionModel-aware routing on the inference gatewayInference
KAI SchedulerGPU placement, and each tenant's GPU guarantee and capKAI Scheduler
NVIDIA GPU OperatorGPU drivers, device plugin, and runtime class on GPU nodesEnable GPU for AI components

Identity, tenancy, and secrets

ComponentRoleDescribed in
KeycloakSign-in, federation, and one Keycloak Organization per tenantKeycloak
BSQAI Tenant OperatorProvisions each tenant's namespace, quota, network policies, identity, storage, database, private MLflow, and GPU queuesBSQAI Tenant Operator
OpenFGARelationship-based authorizationOpenFGA
OpenBaoSecret storeOpenBao
External SecretsSyncs secrets from the secret store into KubernetesExternal Secrets
Sealed SecretsEncrypts secrets kept in GitSealed Secrets
OAuth2 ProxySigns users in to Kubeflow through KeycloakOAuth2 Proxy

Networking and certificates

ComponentRoleDescribed in
Gateway API (Envoy Gateway)Public ingress and HTTP routingGateway API
cert-managerIssues and renews certificatescert-manager
trust-managerDistributes trusted CA bundlestrust-manager
ExternalDNSPublishes DNS records with a configured DNS providerExternalDNS
ReflectorCopies selected Secrets and ConfigMaps across namespacesReflector
Platform network policiesAllow policies for platform namespaces, and an optional platform-wide default denyNetwork policies
CalicoPolicy-only engine that enforces network policies on a cluster without one. Optional, installed at bootstrap.Calico

Observability and security scanning

ComponentRoleDescribed in
GrafanaDashboards for platform operators, with one Grafana Organization per tenantGrafana
PrometheusMetrics collectionPrometheus
Grafana MimirLong-term, tenant-scoped metrics storageGrafana Mimir
Grafana LokiLog storageGrafana Loki
Grafana TempoTrace storageGrafana Tempo
Grafana Alloy and the Alloy gatewayTelemetry collection, and tenant-aware routing of logs and tracesGrafana Alloy
Telemetry query gatewayValidates each telemetry query and restricts it to the caller's tenantTrusted tenant routing
KubescapeDaily configuration and vulnerability scans of running workloads. Optional.Kubescape

Storage, databases, and workflows

ComponentRoleDescribed in
PostgreSQL (CloudNativePG)Relational databases for platform services and tenantsPostgreSQL
PgAdminPostgreSQL administrationPgAdmin
Rook CephIn-cluster S3-compatible object storage, and block and file storage where the site uses Ceph for themRook Ceph
Ceph CSI driversStorage drivers for Ceph volumesCeph CSI Drivers
TemporalDurable, long-running workflows of the BSQAI API, such as file ingestion and reindexingTemporal
Argo WorkflowsMulti-step jobs, pipelines, and batch processesArgo Workflows
Argo EventsEvent-driven automation triggersArgo Events

Delivery and policy

ComponentRoleDescribed in
Argo CDGitOps delivery of every platform applicationArgo CD
KyvernoAdmission policies and generated profile resourcesKyverno
Platform Common CRDsShared custom resource definitionsPlatform Common CRDs
Kata ContainersVirtual-machine isolation of agent sandbox workspaces. Optional, with the agent sandbox.Kata Containers

External integrations

BullSequana AI works with a configured OCI registry and Git repository. It does not deploy them.

IntegrationRoleDescribed in
OCI registryDelivers platform images and chartsHarbor
Git repositoryHolds the GitOps manifestsGitLab

On this page