Healthcare Data Infrastructure Platform
Research and design of a healthcare-native data transformation platform built to help organisations ingest, understand, map, transform, validate, secure and move complex healthcare information.
The challenge wasn't simply to visualise data — it was to make technically complex healthcare data operations understandable, manageable and repeatable for both technical and operational users.
Client identity withheld due to confidentiality.
Context
Healthcare organisations operate with information spread across clinical and insurance systems, legacy platforms, EDI files, HL7 messages, nested JSON, flat files, spreadsheets, custom schemas and FHIR environments. The challenge isn't only moving data — teams must understand source structures, map fields correctly, validate quality, protect personally identifiable information, remove duplicates, convert between standards, monitor execution and maintain traceability. Traditional data-engineering tools are powerful but hard for operational teams; the goal was enterprise-grade workflows without unnecessary complexity.
The problem
Fragmented data sources
Information arrives from many systems and file types.
Manual mapping
Field mapping between systems is repetitive, slow and error-prone.
Poor data visibility
Teams may not understand quality or structure until late in the workflow.
Privacy requirements
Sensitive health information must be handled carefully.
Duplicate records
Repeated records create operational and analytical problems.
Interoperability & operations
Data often needs converting to standards like FHIR, while teams manage projects, pipelines, permissions and settings in one place.
Users
Data engineers
Detailed control over connectors, schemas, mapping, transformation and execution.
Healthcare data analysts
Visibility into structure, quality, anomalies and profiling results.
Implementation specialists
Configure data workflows for new clients, projects or environments.
Project managers
Visibility into project status, ownership and progress.
Administrators
Manage teams, roles, permissions, settings and multi-tenant environments.
Compliance & security
Confidence around privacy, de-identification and controlled access.
Core product modules
Projects
The primary workspace — data sources, connectors, schemas, mapping, pipelines, team, status, logs and outputs.
Connectors
Connect to sources and environments — source selection, credentials, testing, config, status and reuse.
Layouts & sublayouts
Represent incoming/outgoing data structure, including nested, repeating, segmented and hierarchical records.
Mapper
Connect source fields to target fields — transformation rules, status, AI suggestions, validation and conflict resolution.
Profiler
Understand data before transformation — types, nulls, invalid values, distribution, uniqueness, outliers, quality and duplicates.
Pipelines
The end-to-end flow: ingestion → profiling → mapping → transformation → de-identification → deduplication → validation → writing → loading.
Writer & loading
Output format, destination, write rules, validation, status, delivery and failure handling.
Dashboard, team & admin
Active projects, pipeline status, failed runs and quality; users, teams, roles, permissions and multi-tenant settings.
Key capabilities
AI-driven auto-mapping
AI suggests mappings between source and target fields to reduce manual effort while keeping users in control: connect → detect schema → suggest → review → resolve → approve.
Profiling & quality
Identify quality issues before building or running pipelines.
De-identification & re-identification
Remove or mask identifiers; authorised users can restore identity where permitted, preserving integrity and governance.
Deduplication & FHIR conversion
Identify and remove repeated records; transform incoming data into healthcare interoperability formats.
Key user journey
- Create project
- Add data source
- Configure connector
- Detect layout
- Profile data
- Map fields
- Configure transformations
- Add privacy & quality rules
- Build pipeline
- Test → run → monitor
- Write & load output
Design approach
Visualise the workflow
Show clearly where data comes from, what happens to it and where it goes.
Progressive technical detail
Basic users shouldn't be forced into advanced configuration immediately.
Actionable validation
Warnings and errors should be clear, traceable and tied to the right field or stage.
Separate setup from monitoring
Creating a pipeline and monitoring a running one are different jobs.
Reusable patterns
Consistent tables, filters, forms, drawers, statuses, logs and empty/error states.
Preserve human control
AI suggestions should be reviewable and editable, not automatic authority.
Major product decisions
- Projects as the primary container
- Source & target visible during mapping
- AI as suggestion, not authority
- Data quality surfaced before transformation
- Pipeline status understandable at a glance
- Privacy workflows as first-class features
- Support technical & operational users
- Multi-tenancy in the foundation
- Configuration, execution & monitoring separated
Product complexity
- Healthcare-specific workflows
- Sensitive information
- Complex, nested data structures
- Multiple source formats
- Multi-stage pipelines
- Low-code interaction design
- Role-based permissions
- Multi-tenant architecture
- Data quality, privacy & interoperability
What this project demonstrates
Selected project details have been anonymized due to client confidentiality. No client names, logos or private information are shown.
More work
All work →Building a complex product?
I work with founders, agencies and businesses that need help turning complex ideas, workflows and systems into clear digital products.
