Skip to content
TETRALITICS

Technology

Designed for Google Cloud

TETRALITICS is designed around a Google Cloud-native target architecture. The current site is a browser-based prototype; the architecture below describes our planned production design, not services that are already deployed.

Planned Production ArchitectureTarget design · not currently deployedPlanned

  1. Google Identity

    Institutional sign-in · Workspace SSO

  2. Application Services

    Cloud Run · stateless containerised services

  3. Socratic Learning Engine

    Vertex AI · Gemini models · pedagogical guardrails

  4. xAPI Event Processing

    Pub/Sub · validated learning statements

  5. Data Platform

    Cloud SQL (operational) · BigQuery (analytical)

  6. Learning Analytics & Competency Intelligence

    Dashboards · competency profiles · evidence

Architecture principles

Responsible by design, built to scale with institutions.

These principles guide our planned production platform. They describe intended design, not current certifications or deployed services.

  • Google Cloud-native target architecture

    Managed, serverless services — Cloud Run, Pub/Sub, Cloud SQL and BigQuery — chosen to scale with institutional demand while limiting operational overhead.

  • Gemini and Vertex AI integration plans

    We plan to build the Socratic Learning Engine on Vertex AI with Gemini models, wrapped in pedagogical prompting, evaluation and guardrail layers.

  • Google Workspace SSO interoperability

    Planned single sign-on through Google identity so institutions using Google Workspace or Workspace for Education can onboard without new credentials.

  • xAPI learning event ingestion

    Learning activity is modelled as xAPI statements, published to Pub/Sub, validated and stored for analytics and competency evidence.

  • Multi-tenant institutional architecture

    Logical tenant isolation per institution, with tenant-scoped data access, configuration and administration.

  • GDPR-aware data lifecycle

    Designed around data minimisation, purpose limitation and defined retention periods. Formal compliance will be assessed with institutional partners before production use.

  • EU data residency strategy

    Our strategy is to host production learner data in EU Google Cloud regions, supporting European institutions’ residency expectations.

  • Data deletion workflows

    Planned workflows for learner and institution-initiated deletion, propagated across operational and analytical stores.

  • Human oversight for educational AI

    Educators can review AI interactions, adjust scaffolding and override outcomes. AI supports — but does not replace — professional judgement.