SureCentric Tech Architecture

SureCentric's architecture is organized into five layers: the desktop application layer, the dynamic data warehouse, the schema layer, the AI services layer, and the data sources layer. Each layer has a clearly defined contract with the layers above and below it.

Architecture Diagram

Layer Breakdown

Desktop Layer — UI Surface

The desktop layer provides the user-facing surfaces for all platform operations:

Surface	Description
Dashboards (Superset)	Apache Superset embedded as an iframe — charts, SQL Lab, datasets
Query & Reports	Structured query execution with schema-grounded NL2SQL via DataAgent
Explorer & Viewer	Browse schemas, data, and ontology relationships interactively
Schema Builder	JSON-LD schema editor — the primary authoring and translation tool

Data Warehouse Layer

The lakehouse sits beneath the desktop layer and executes all analytical queries. Schema DDL is generated automatically from the JSON-LD schemas — tables, views, indexes, foreign keys, and constraints are never hand-authored.

Engine	Purpose
DuckDB	In-process, Arrow-native analytical engine for local and embedded execution
Dolt (MySQL wire)	Version-controlled metadata store for platform data, CARD profiles, and relational schemas — supports Git-style branching, commits, diffs, and rollback. Replaced PostgreSQL.
Apache Iceberg	Time-travel, schema evolution, and immutable data snapshots
Trino	Federated SQL across heterogeneous data sources

note

SureCentric migrated its metadata database from PostgreSQL to Dolt (MySQL-compatible). See Dolt Database for the full architecture and branching strategy.

Schema Layer

The schema layer is SureCentric's core innovation. JSON-LD is the canonical format — every data source, every query, and every dashboard is grounded in a declared, versioned, semantic schema.

┌─────────────────────────────────────────────────────────────────────┐
│  Schema Layer                                                        │
│                                                                      │
│  ┌─────────────┐    ┌────────────────┐    ┌────────────────────┐   │
│  │ Schema      │    │ Schema         │    │ Exporters          │   │
│  │ Builder     │───►│ Mappers +      │───►│ • OWL / RDF-XML    │   │
│  │ (Editor)    │    │ Translators    │    │ • SQL DDL          │   │
│  └─────────────┘    └────────────────┘    │ • JSON Schema      │   │
│                                           └────────────────────┘   │
└─────────────────────────────────────────────────────────────────────┘

AI Layer

Component	Role
SureAgent	Agentic workflow orchestration — multi-step AI pipelines
SureIntel	Flowise-based RAG pipeline for document intelligence and retrieval
SureLLM	Private domain LLM gateway (LiteLLM-compatible, OpenAI API surface)
SureIndex	Universal data ingestion via CocoIndex — indexes any structured or unstructured source
DataAgent	NL2SQL engine grounded by compiled schema metadata — deterministic, not hallucinated

Data Sources Layer

Source Type	Examples
SureDrive Apps	SureDrive document archives with OWL-based content models (CM1, CM2, CM3)
External Databases	PostgreSQL, MySQL, Snowflake, BigQuery, Oracle
SureConnect	OpenAPI 3.x connector for any REST API
File Sources	Parquet, CSV, JSON-LD, XLSX

Schema Builder Workflow

① JSON-LD Editor (Schema Builder)
   • Create/edit schemas with @context, @type, @id
   • SPARQL autocompletion against ontology endpoints
   • Visual schema diagram (Vue Flow)
   • AJV validation in real-time

            ↓

② JSON-LD Canonical Schema
   • CareLex/eClinical @context namespace
   • OWL import/export for SureClinical archives
   • Schema catalog/registry with versioning

            ↓

③ Schema Translation Service
   • JSON-LD → DDL (PostgreSQL/DuckDB CREATE TABLE)
   • JSON-LD → OWL (Nuxeo archive format)
   • JSON-LD → JSON Schema (API validation)

            ↓

④ Data Warehouse / Lakehouse
   • Tables, views, metrics auto-created from DDL
   • Superset datasources/datasets auto-registered
   • NL2SQL grounded by compiled schema metadata

Tech Stack

Category	Technology
Schema editor	Vue 3, TypeScript, PrimeVue 4, Pinia, Vite, TailwindCSS
Schema format	JSON-LD 1.1 (W3C), OWL/RDF-XML, JSON Schema Draft 2020-12
Validation	AJV (JSON Schema), jsonld.js (JSON-LD 1.1)
Visualization	Vue Flow (schema diagram), Apache Superset (dashboards)
Analytical engine	DuckDB, Dolt/MySQL, Apache Iceberg, Trino
AI gateway	LiteLLM (OpenAI-compatible)
OWL service	Spring Boot + owlapi-jsonld
Deployment	Docker Compose, Cloudflare Pages (docs), Cloudflare Workers (MCP index)