We work natively within Foundry Code Repositories — building sophisticated pipelines, not bolt-ons.

• Magritte framework for all EPR connections (Epic, Oracle Health, System C, InterSystems) with incremental sync using high-water mark logic
• Snapshotting and partitioning for reference data with automated schema validation at point of entry
• Build halts automatically if source schema deviates — preventing downstream Ontology corruption

A disciplined domain modelling approach that prevents duplicate objects and ensures a single view of the patient.

• Core Objects (Patient, Staff Member, Location, Clinical Event) - centrally governed, changes require Design Authority approval
• Domain Objects (e.g., Cancer_Pathway_Milestone, Theatre_Kit_Usage) - use-case specific, linked to Core entities
• "Extend, Don't Duplicate" philosophy - existing objects are extended before new ones are built
• Objective: 80–90% of new analytical questions answered through Ontology traversal, not new engineering

A Data Twin must be interactive - not just a read-only reporting layer.

• Clinical and operational users act on data directly - validating waiting list entries, assigning cancellation reasons, flagging patients for expediting
• Actions trigger controlled side-effects - notifications, audit log entries, and status updates across linked objects
• Phonograph Sync ensures edits are reflected instantly in Workshop while being committed asynchronously to the data layer

Structured governance to keep the Data Twin accurate, lean, and performant as the platform matures.

• Ontology Registry - all approved Object Types and Link Types maintained in a central register
• Joint Trust/VE3 Design Authority meeting bi-weekly to govern Core Ontology changes
• Object Deprecation Strategy - a lifecycle process for retiring obsolete objects and preventing technical debt

A deliberate Foundry + Azure model that maximises the strengths of both platforms.

• Foundry hosts the Ontology, decision logic, and operational applications - Compute Credits preserved for interactive workloads
• Azure handles high-intensity tasks: genomics sequencing, DICOM processing, Monte Carlo simulations, large-scale stratification
• Secure bridge mechanism passes raw data to Azure and returns only high-value outputs (risk scores, model results) back to Foundry
• Strict Raw → Curated → Semantic three-tier architecture ensuring auditability, reusability, and clinical safety

As the national provider of NHS-PET, privacy is embedded into the pipeline - not applied as an afterthought.

• Format Preserving Encryption (FPE) for consistent pseudonymisation enabling longitudinal analysis
• SHA-256 salted hashing for strictly one-way, irreversible flows
• NLP-based Leaky PID scanning - free-text fields scanned during ingestion, identifiable data automatically redacted or quarantined
• Foundry Markings and Row-Level Policies ensuring users only access data aligned with their consent and purpose

Quality gates embedded directly into pipeline code - not monitored after the fact.

• Referential integrity checks - orphaned records halt the build automatically
• Statistical drift monitoring - alerts triggered if ingestion volume drops more than 20% against the 30-day average
• Zero-tolerance null checks on critical fields (NHS Number, Event Date)
• Lineage impact analysis before every code change - downstream dependencies mapped to prevent silent breakages

Built to grow your team's capability - not your dependency on us.

• Pod-based co-development: Foundry Solution Architect, Engineers, Clinical Data Architect, and embedded Trust counterparts working alongside your teams
• Knowledge transfer embedded into delivery KPIs - your team progressively assumes ownership
• Bespoke training for every user group - from Data Literacy for operational staff to advanced Foundry engineering for informatics teams
• Innovation Sandbox for safely piloting AI models and emerging technologies before live deployment

A deliberate Foundry + Azure model that maximises the strengths of both platforms.

• P1 (<15 mins), P2 (<1 hour), P3 (<4 hours) - incident response tiered by clinical impact
• Proactive pipeline latency monitoring and staleness checks - issues identified before clinicians notice
• Strict Dev → Staging → Production environments with mandatory peer review - we never code in production
• Dedicated Service Delivery Manager as your single point of contact for all oversight and escalation