Texas Department of Motor Vehicles · ITSD Division

DATA ARCHITECT I
INTERVIEW COMMAND CENTER v2

Interview Confirmed Tue Apr 21 · 11:00 AM $8,500–$9,167 / mo Remote · Texas

// Updated Strategic Brief — Panel Intel Received

You're interviewing with Michael Demaree (CDO) — your direct boss — and Terry Cowart (Enterprise Architect) — the standards and governance voice. Demaree came up through 9 years of hands-on data architecture at GM Financial and used Erwin Data Modeler daily. Cowart has a patent in ontological data modeling, mainframe/ADABAS depth, and 6+ years shaping TxDMV's enterprise architecture. These are serious technical peers, not HR screeners. Calibrate accordingly: speak architecture, not job-search.

Panel Intelligence

Michael Demaree

Chief Data Officer · Your Direct Boss

TxDMV ITSD · Waco, TX · CDO since Dec 2025

20 Yrs IT 9 Yrs Data Architecture AZ-900 Certified ITIL Foundation Enterprise Design Thinking

Background

▸9 years as Data Architect/Modeler at GM Financial (OLTP + OLAP, full lifecycle)

▸Mainframe Developer at Fidelity Investments (2011–2014) — production data & accounts

▸Joined TxDMV May 2024 as Data Architect; promoted to CDO December 2025

▸Military background: Air Force / L-3 Communications Program Manager

Tools & Tech He Lives In

▸Erwin Data Modeler — his primary data modeling tool (listed in top skills)

▸TIBCO Data Virtualization, Azure DevOps, Oracle Exadata, TOAD

▸SQL, Python, COBOL, JCL, Java — broad stack

▸Tableau, Salesforce, Mainframe, ServiceNow

How to Win Demaree

He built data warehouses from scratch for a decade. Speak the same language: data lineage, storage optimization, performance tuning, full lifecycle ownership. He knows when someone is faking architecture depth. Acknowledge Erwin directly — say you're tool-agnostic but have modeled the concepts Erwin manages. Mention your OLTP/OLAP dimensional modeling work at JPMorgan and Bank of Hawaii. He'll recognize it.

Terry Cowart

Enterprise Architect · Standards & Governance Voice

TxDMV ITSD · Austin, TX · EA since Feb 2020

6+ Yrs TxDMV US Patent Holder ITIL v3 Python for ML Statistics for Data Science

Background

▸Patent: "Systems and Methods for Ontological and Meta-Ontological Data Modeling" (2013) — currently extending toward ML/AI and predictive analytics

▸NTT DATA: Director / Enterprise Architect for TxDOT, TWC (ADABAS/Natural), TxDMV (6+ yrs pre-hire)

▸Texas.gov: Lead Architect of payment gateway processing billions in annual revenue

▸Unisys: Chief Architect for HOLMES (Scotland Yard), Louisiana DMV — public sector pedigree

What He Cares About

▸SDLC discipline — how your architecture decisions fit the full delivery lifecycle

▸Enterprise integration — how your data layer connects to the broader system landscape

▸Standards and governance — naming conventions, metadata, ontology, consistency

▸AI/ML direction — his patent evolution signals he's watching this space actively

How to Win Cowart

He thinks at the ontological level — data as a conceptual model, not just tables. When discussing governance and standards, go beyond "we named things consistently" — talk about data ownership, lineage, metadata contracts, and how data models serve as the enterprise truth source. Mention your Anthropic AI certifications when he opens the AI door — his patent is already moving toward ML. He'll respect a peer who thinks in systems, not solutions.

Alignment Scorecard

Overall Fit

94%

Exceptional match

Required Exp.

15 YRS

Min req: 5 years

SQL / DB2 / SP

100%

All three stacks

ETL Tools

85%

ADF ✓, Informatica gap

Cloud Coverage

3/3

AWS · Azure · GCP

Erwin / ER Studio

Tool-agnostic; bridge it

Gov / Compliance

✓

SOX, HIPAA, audit

AI / Anthropic

★★★

Certified · Differentiator

Strongest Cards

Play Hard

Dimensional Modeling — Kimball / Inmon / full lifecycle100%

SQL Server / DB2 / Stored Procedures (15 yrs)100%

OLTP + OLAP — JPMorgan, Bank of Hawaii (2M+ tx/day)100%

ETL / Pipeline Architecture (ADF, Airflow, DBT, Glue)97%

Data Governance & Metadata Standards95%

Stakeholder Engagement — exec to end-user (15 yrs)100%

Azure (certified) + AWS + GCP tri-cloud92%

AI / Anthropic Certified (Claude, AI Fluency, Code)90%

Gaps — Scripted Bridges

Prep These

📐

Erwin Data Modeler (Demaree's tool)

"I haven't used Erwin directly, but I've designed the conceptual, logical, and physical models that Erwin manages — in Visio, DbSchema, and ERD tools. Erwin is the interface; the modeling craft is what I bring. The learning curve is the UI, not the discipline."

⚡

Informatica PowerCenter / Cloud ETL

"Six years of Azure Data Factory covers the same enterprise ETL orchestration patterns — parameterized pipelines, lineage, monitoring, and governance. Informatica patterns transfer directly; I'd be productive within the first sprint."

🖥️

Mainframe / ADABAS / Spring Batch (Cowart's domain)

"I've worked with DB2 mainframe integrations at JPMorgan and Bank of Hawaii — I understand the batch paradigm, JCL concepts, and mainframe data access patterns even without direct ADABAS hands-on. The integration layer is familiar territory."

✅

Public Sector / Government Data (Strength)

Alamo Colleges is a Texas public institution — 60,000 students, 5 campuses, compliance reporting, institutional governance. Directly analogous to state agency data responsibility. Lead with this when context of public sector comes up.

Question Playbook — Calibrated to Panel

● Demaree (CDO) ● Cowart (EA) ● Either

T-01 Demaree

Walk us through how you approach designing a data warehouse from scratch.

"Business-first, then model — I always start with the questions leadership needs to answer, not the tables that exist."

I begin with stakeholder interviews to map what decisions the data needs to support. From there I do a source system inventory, identify systems of record, and build a conceptual model. Then the dimensional model — star schemas for aggregated analytics, normalized staging for operational data. At JPMorgan Chase I architected the CDR warehouse integrating Oracle, SQL Server, and cloud sources processing 2M+ daily transactions. At Alamo Colleges I built ETL workflows integrating Banner, Colleague, and Canvas into a unified institutional reporting warehouse. I always design for three things simultaneously: performance, governance, and extensibility.

Star SchemaKimballInmonSystems of RecordOLTP StagingFull Lifecycle

T-02 Demaree

Describe your DB2, SQL Server, and stored procedure experience.

"This is my daily language for 15 years — I can speak to all three with production war stories."

SQL Server is my primary RDBMS — 15+ years at UWM, JPMorgan, and in consulting. I've written complex T-SQL stored procedures for data transformation, parameterized reporting, and ETL automation. I tuned queries that dropped report generation from 8+ hours to 45 minutes through indexing, query plan analysis, and partitioning. DB2 at JPMorgan Chase and Bank of Hawaii for mainframe integrations — extraction, reconciliation, and legacy interface work. Stored procedures are a core part of my toolkit for encapsulating business logic and building reusable ETL components.

T-SQLStored ProceduresDB2 MainframeQuery TuningIndexingPartitioning

T-03 Either

We use Informatica. What's your ETL tool experience?

"My ETL depth is in Azure Data Factory and Apache Airflow — the enterprise patterns are identical."

Six years of production Azure Data Factory — pipeline orchestration, parameterized datasets, trigger scheduling, monitoring, lineage, and error handling. Plus Apache Airflow for complex DAG-based workflow orchestration and DBT for version-controlled transformation layers. I haven't sat in Informatica's UI, but the conceptual ETL model — extract, validate, transform, load, reconcile — is the same across all enterprise tools. I'd be productive in Informatica within the first sprint; the patterns transfer directly.

Azure Data FactoryApache AirflowDBTAWS GluePipeline Orchestration

M-01 Demaree

We use Erwin Data Modeler. What data modeling tools have you used?

"I'm tool-agnostic, but deeply methodological — I've designed everything Erwin manages."

I haven't used Erwin directly, but I've built the conceptual, logical, and physical data models that Erwin is designed to manage — using Visio, DbSchema, ERD tools, and draw.io. I'm fully versed in Erwin's modeling capabilities: entity-relationship notation, physical model generation, forward/reverse engineering, and metadata management. Erwin is the interface; the data modeling craft is what I bring, and that craft transfers in hours, not weeks. I'd welcome the chance to get hands-on with Erwin's environment in week one.

Erwin BridgeERDConceptual / Logical / PhysicalForward EngineeringMetadata

M-02 Demaree

How do you ensure data models perform at scale in an enterprise data warehouse?

"Performance is designed in, not tuned out — I build with execution plans in mind from the first schema draft."

I design for read patterns first in OLAP systems — identifying the most frequent query shapes and optimizing fact table grain and dimension keys accordingly. Then I apply indexing strategies (clustered on most selective keys, covering indexes for frequent joins), partitioning on date or entity dimensions, and columnar storage where the platform supports it. At JPMorgan I achieved 50% query performance improvement on a warehouse handling millions of daily transactions. In Snowflake-based consulting work I've achieved sub-100ms query performance through clustering keys and micro-partition optimization. I also instrument pipelines to catch schema drift and volume anomalies before they become performance incidents.

Query OptimizationPartitioningClustering KeysColumnar StorageSchema Drift

M-03 Cowart

How do you approach master data management and data quality at the architecture level?

"MDM starts at the model — if the architecture doesn't enforce a single version of truth, no downstream tool can fix it."

I approach MDM by first establishing systems of record — formally designating which source system owns each master entity (customer, vehicle, registration). The data model then enforces that designation through foreign key relationships and surrogate key strategies. I implement data quality rules at the ETL ingestion layer — completeness, conformance, uniqueness, and referential integrity checks — with rejection queues and reconciliation reports. At JPMorgan, I implemented automated validation frameworks that reduced incident response time by 70%. Data quality isn't a cleanup job — it's an architecture property.

Master Data ManagementSystems of RecordSurrogate KeysData Quality RulesReferential Integrity

G-01 Cowart

How have you established data standards and governance frameworks in large organizations?

"I've built governance from scratch twice — once for 1,700 engineers at UWM, and once across five campuses at Alamo Colleges."

At UWM I started with a current-state audit — what patterns were causing production incidents, what the business needed, what didn't exist. I drafted SQL development standards collaboratively with senior engineers (not top-down), piloted with one team, incorporated feedback, then rolled out with mandatory training. Standards included naming conventions, indexing requirements, stored procedure templates, ETL error handling patterns, and documentation standards. At Alamo Colleges I defined data governance across five campuses: ownership models, data quality thresholds, and metadata contracts for institutional reporting. Governance that isn't co-created doesn't get followed.

Data StandardsNaming ConventionsData OwnershipMetadata ContractsChange Management

G-02 Cowart

How do you configure and establish a data catalog for enterprise data management?

"A data catalog is only as good as the processes that feed it — tooling is secondary to taxonomy design."

I start by defining the metadata taxonomy: what entities need cataloging, who owns them, what lineage needs to be tracked, and what the governance trigger events are (new source system, schema change, data classification change). At Alamo Colleges I built the conceptual framework for cataloging academic and administrative data across five systems. In consulting work I've designed technical metadata schemas, automated lineage capture through ETL instrumentation, and data quality dashboards. For TxDMV I'd approach the catalog as a living governance artifact — not a documentation exercise — with automated feeds from ETL pipelines and scheduled metadata reconciliation.

Data CatalogMetadata TaxonomyData LineageAutomated FeedsGovernance Triggers

G-03 Either

What's your experience with regulatory compliance in data environments?

"I've designed in SOX, HIPAA, and SOC 2 environments — compliance is baked into the data model, not bolted on after."

At JPMorgan Chase, every architecture decision ran through SOX audit requirements — data lineage, access logging, separation of duties, and immutable audit trails were architecture properties, not features. In healthcare consulting I designed HIPAA-compliant systems with field-level encryption, RBAC, and row-level audit logging. For TxDMV, I'd apply the same principle: state and federal data governance regulations get built into the model from day one. Data minimization, retention schedules, access classification — these are architecture decisions, not IT policy footnotes.

SOXHIPAAAudit TrailsRBACData RetentionData Classification

B-01 Either

Tell me about a time you led a complex data initiative with competing priorities.

"At UWM I managed 8 concurrent enterprise training programs for 1,700 engineers while simultaneously establishing SQL standards the entire org had to adopt."

I prioritized by downstream impact — which teams were blocked waiting for standards, which programs had compliance deadlines, which had the largest ripple effect if delayed. I established a weekly stakeholder sync cadence, created a shared project tracker, and delegated delivery of three programs to senior engineers I'd mentored into internal trainers. Result: All 8 programs delivered on schedule. Developer onboarding dropped from 3 months to 6 weeks. Production stability improved 40%. The key was treating competing priorities as a systems problem, not a scheduling problem.

STAR MethodPrioritizationDelegationStakeholder SyncImpact Mapping

B-02 Demaree

Describe a significant data quality or pipeline issue you identified and resolved.

"At JPMorgan, our CDR reconciliation pipeline was producing silent discrepancies — the kind no alert catches until someone runs the quarterly numbers."

P&L figures weren't reconciling between Oracle source and the SQL Server warehouse. I traced it to a stored procedure applying currency conversion after aggregation instead of before — a subtle ordering bug with significant financial impact. The fix was straightforward, but more importantly, I implemented automated reconciliation checkpoints at each ETL stage and a data quality dashboard that catches anomalies within hours, not quarters. Root cause methodology: trace from output back to source, isolate the transformation step, validate against raw data. That methodology is now standard in all my pipeline designs.

Root Cause AnalysisReconciliationAutomated ValidationData QualityAnomaly Detection

B-03 Cowart

How do you handle conflicts between business requirements and architecture best practices?

"I treat it as a risk conversation, not a disagreement — my job is to make the tradeoffs visible."

At Alamo Colleges, academic stakeholders wanted denormalized, duplicated data for faster reporting — it was faster to query but would drift out of sync. I'd present the business requirement alongside two options: the quick path with its technical debt implications, and the architecturally sound path with a timeline. I never just say no — I make the cost of each choice tangible. When leaders understand that the shortcut creates a data reliability problem they'll own in 18 months, they usually choose correctly. When they choose speed, I document the decision and build a migration path into the roadmap.

Technical DebtOptions FramingRisk CommunicationArchitecture GovernanceRoadmap

A-01 Demaree

What's your cloud platform experience?

"Certified on Azure, production experience across all three clouds — and I think about cloud architecture with state agency constraints in mind."

Azure: Deepest — certified (DP-900, AZ-900), production use of Data Factory, Synapse Analytics, Functions, Blob Storage, and Azure SQL. AWS: S3, Lambda, Glue, Redshift, Athena for data engineering workloads. GCP: BigQuery for large-scale analytics. For a state agency, my relevant considerations are data residency (Texas law may require in-state hosting), security certifications (StateRAMP), and portable data formats to avoid vendor lock-in. I'd favor open formats and platform-agnostic ETL patterns so TxDMV retains strategic flexibility.

Azure CertifiedMulti-CloudStateRAMPData ResidencyOpen Formats

A-02 Cowart

The JD mentions AI technologies. What's your AI implementation experience?

"This is a genuine differentiator — I'm Anthropic-certified across multiple tracks and actively building AI-integrated data systems."

I hold multiple Anthropic AI certifications: Claude 101, Claude Code in Action, AI Fluency Framework, AI Fluency for Educators, Teaching the AI Fluency Framework. In consulting I've integrated LLM-based components into data pipelines for automated anomaly detection, intelligent data classification, and natural language query interfaces over structured datasets. For government data systems the practical AI applications I'd prioritize at TxDMV: automated data quality flagging, intelligent metadata tagging for the data catalog, anomaly detection in DMV transaction data, and NL access to institutional data for non-technical staff. AI in government data architecture is about augmenting data trust and accessibility — not replacing governance.

Anthropic CertifiedLLM IntegrationAnomaly DetectionNL QueryAI Governance

Confirmed Logistics

Interview Details

📅

Date & Time

Tuesday, April 21, 2026 · 11:00 AM CST

🎥

Format

Microsoft Teams · Camera On · ~60 min

🔗

Join Link

teams.microsoft.com/meet/21930527183805

🔑

Meeting ID / Passcode

219 305 271 838 05 · 8a7Db6ZQ

📞

Dial-In (Austin)

+1 737-787-8456 · Conf ID: 356 421 622#

👥

Panel

Michael Demaree (CDO) · Terry Cowart (EA)

🎓

Graduation Note

Apr 15–16 → full weekend to prep before interview

Pre-Interview Checklist

✅

Done

Email replied · Calendar invite accepted

📋

Research

Review Erwin Data Modeler capabilities (1hr)

📋

Research

Review Demaree's LinkedIn / GM Financial work (30 min)

📋

Research

Skim Cowart's patent abstract (15 min)

📋

Prep

Rehearse T-01, M-01, G-02 answers (core likely questions)

📋

Tech Check

Teams test call · Camera · Lighting · Background

📋

Day-Of

Join 5 min early · Anthropic cert tab open · JD nearby

Closing Strategy

Mirror Demaree's language. He's a data modeler at heart. Use "data lineage," "storage optimization," "performance tuning," and "full lifecycle" naturally — not as buzzwords, as the actual substance of your answers.

Engage Cowart's systems thinking. When he pushes on enterprise integration, talk about how your data architecture plugs into the broader system landscape — APIs, event streams, MDM, and how metadata governance spans beyond the data layer.

Drop the Anthropic credential explicitly. The JD calls out AI knowledge. Cowart's patent is evolving toward ML. Neither of them will have an Anthropic-certified candidate often. Name it — don't bury it.

Close with a data catalog insight. The JD lists configuring and establishing a data catalog as a core duty. Leave them with a specific framework thought — shows you've already scoped their problem before day one.

Questions to Ask Them

Always Ask 3

→

Michael, what does the current data architecture look like at TxDMV, and what are the biggest gaps this role is being hired to close?

Direct to CDO. Shows you're thinking about the real problem, not the job posting. Will tell you everything you need to know about the first 90 days.

→

Terry, where is TxDMV in its data catalog and metadata management journey — is this a build, a rebuild, or a maturation effort?

Direct to Enterprise Architect. Core to the JD. Shows you read the role carefully and are already thinking in implementation terms.

→

What does success look like for this role at the 90-day and 6-month marks?

Anchors in outcomes. Shows you're thinking about impact, not onboarding.

→

Terry, is there an active AI or ML initiative within ITSD, or is that an area TxDMV is still defining its strategy for?

Opens the Anthropic certification conversation naturally. Cowart's patent evolution makes this especially relevant to him.

→

What's TxDMV's cloud strategy — primarily Azure, or multi-cloud? Any StateRAMP or data residency constraints shaping platform decisions?

Practical, specific, and signals you understand government cloud constraints. Demaree will appreciate the operational detail.

    Vi2sDk · TxDMV Data Architect I · Job #00056033 · Command Center v2 · Panel Intel: Demaree + Cowart · Updated Apr 10, 2026
  