Gen AI CoE Services Ecosystem

Overview

The Gen AI Center of Excellence (CoE) at Cambridge University Press & Assessment needed more than just a set of tools — it needed a coherent platform that could scale AI adoption across non-technical teams while maintaining governance and quality standards.

I led the vision, architecture, and delivery of the CoE Services Ecosystem: a suite of internal AI services, internal APIs, and governance tooling that forms the backbone of CUP&A's AI operations.

What we built

• Evaluations Service — AI system quality gate for all Cambridge AI systems pre-production
• Documents Service — Centralized document parsing, indexing, and management platform with RAG capabilities (cached vectors, knowledge graphs)
• Operation Automations — Automating day-to-day tasks: FAQ management, email response integrations, systems monitoring
• RAG Pipeline Asset: Pilot — Centralized RAG solution for non-tech teams that want a chatbot that knows their documents
• Lesson Planner + Assessments — Educational content generation with filled lesson planner PDF output

Architecture principles

The ecosystem is designed around a few hard constraints: it must be understandable by non-technical stakeholders, it must be auditable, and it must not create vendor lock-in.

Each service is independently deployable, with well-defined APIs and fallback paths. Governance tooling ensures every AI output passes a quality gate before reaching production.

Status

Active development. Several services are in production; others are in the piloting phase with specific teams.