Domain-driven data
Define shared scientific entities, relationships, system-of-record boundaries, and lineage so the same concept means the same thing as it moves across the ecosystem.
DATA MANAGEMENT ARCHITECT · BIOTECH
Scientific data has a shape.
I architect how scientific data is modeled, connected, governed, and made usable — preserving the experimental context that gives biotech data its meaning.
DATA ENABLEMENT & ANALYTICSData Management Architect
THE WORK NOW
Not “move the data.”
Keep its meaning intact.
My focus is not a single application or database. It is the structure between them: how samples, assays, lots, batches, instruments, methods, and results are represented consistently across lab systems, data platforms, and analytics.
DOMAIN-DRIVEN DATA MANAGEMENT
Start with the scientific domain.
A sample is not just a row. An assay is not just a table. The architecture has to preserve relationships, provenance, ownership, and context across systems.
DOMAIN RELATIONSHIP LINEAGE
SCIENTIFIC
CONTEXT
CONTEXT
Sample
Assay
Lot
Batch
Instrument
Method
Integration patterns
Shape batch and streaming patterns between laboratory platforms, operational systems, central data platforms, curated datasets, and analytical consumers.
Metadata & governance
Make context durable through metadata, reference/master data, quality controls, lifecycle rules, catalog standards, and traceability.
Data products & decisions
Turn governed scientific data into reusable, analysis-ready layers with consistent semantics for visualization, metrics, and downstream use.
WHY BIOTECH IS DIFFERENT
The data is inseparable from the experiment.
In consumer, finance, or conventional enterprise systems, a record can often stand on its own. Scientific data is different: its meaning depends on experimental context, method, material history, instrument conditions, provenance, and the relationships between them.
Context is part of the data
A result without the sample, assay, method, conditions, and provenance that produced it can quickly lose scientific value.
Lineage is scientific memory
Traceability is more than an audit trail. It preserves how materials, methods, transformations, and observations relate across the lifecycle.
The domain comes first
Architecture should reflect how scientists actually work rather than forcing scientific concepts into generic application or database boundaries.
RESULT+SAMPLE+METHOD+CONDITIONS+LINEAGE= SCIENTIFIC MEANING
FROM AUTOMATION TO DATA
I built the systems that generated the data.
Years in laboratory automation gave me a physical understanding of scientific data. Every field has an origin: a sample was moved, a method was executed, an instrument was configured, and a result was produced. Good architecture preserves that context.
01Experiment
→Sample · assay · method · instrument
02Execution
→Automation · scheduling · QC · capture
03Context
→LIMS / ELN · APIs · metadata · lineage
04Architecture
Domain models · governance · analytics
01PHYSICAL
Workflow
Samples
Liquid handling
Instruments
Assays
02DIGITAL
Context
Schedulers
LIMS / ELN
APIs
Metadata
03ARCHITECTURE
Meaning
Domain models
Lineage
Governance
Analytics
ARCHITECTURE LANDSCAPE
Source systems are only the beginning.
This is the conceptual path I work across — not a vendor checklist, but the layers that have to connect.
01
Scientific source
LIMS
ELN
Automation
Instrument data
→02
Integration
REST APIs
ETL / ELT
Streaming
SQL
→03
Data platform
Data lake patterns
Curated datasets
Metadata / catalog
Relational · NoSQL · Graph
→04
Consumption
Semantic layers
Spotfire / Tableau
Metrics
AI-ready context
Current role landscape includes cloud data platforms such as AWS / Snowflake / Databricks-class ecosystems and scalable LIMS / ELN platforms. The emphasis here is architecture and integration patterns rather than claiming a tool inventory.
SELECTED PROOF
Three examples. No résumé dump.
Clinical sample ingestion
Led a workflow that transformed site/CRO clinical sample information into structured, LIMS-ready data with lineage and became a primary daily ingestion mechanism.
Operational observability
Partnered with IT to stream liquid-handler log data into Splunk for monitoring, semantic formatting, archival, troubleshooting, and performance analysis.
Scientific integration
Connected automation, SQL-backed sample tracking, and ELN REST APIs so experiment context and sample lineage could survive system boundaries.
AI & IMPLEMENTATION
The threshold for AI is not access to a model.
The real threshold is understanding the system well enough to design where AI belongs, connect it to trustworthy context, and make its output useful inside a scientific workflow.
MODEL
+
CONTEXT
+
SYSTEM DESIGN
=
IMPLEMENTATION
Understand the domain
AI is only as useful as the context around it. In biotech, that means samples, assays, methods, instruments, provenance, and the scientific relationships between them.
Design the connection
The value appears when models are connected to governed data, APIs, workflows, metadata, and human decision points—not when AI sits in isolation.
Implement with intent
The question is not “Where can we add AI?” It is “Where can AI reduce friction, surface understanding, or improve a decision without losing scientific meaning?”
BACKGROUND
Three lenses on the same problem.
2005 →SCIENCE
Understand the experiment
My foundation is laboratory science and bioengineering. That matters because useful scientific architecture starts with the experiment, not the database.
2008 →AUTOMATION
Encode the workflow
Automation moved the problem from performing experiments to designing repeatable systems — workflows, instrumentation, QC, LIMS integration, and structured outputs.
TodayDATA
Preserve the meaning
The scope is now broader: domain models, lineage, metadata, platform patterns, governance, analytics enablement, roadmaps, and cross-functional delivery.
COMPUTING
M.S. Computer Science
Columbia UniversityMachine Learning · GPA 3.8BREADTH
MBA · Business & strategy
B.S. Bioengineering · University of Washington
Certified SAFe Scrum Master · SSM 6.0
Engineer-in-Training · NCEES
SAFe®
DELIVERY MATTERS
Architecture has to ship.
Certified SAFe Scrum Master (SSM 6.0). I use roadmap thinking, design reviews, PI planning, and iterative delivery to move cross-functional data work from diagrams into operating systems.
ENOCH SHUM