Cohort 4Register

Architecture · Growing

Data Architect: Skills, Projects & Interview Questions (2026)

Design enterprise data architecture, models and governance.

Updated

Demand 7/102026 outlook 8/10Difficulty 8/10High remote

What a Data Architect actually does

Designing enterprise data architecture, models, governance and platforms.

Top hiring companies: Microsoft, Amazon, Databricks, Snowflake, Deloitte, Accenture.

Top industries: Finance, Retail, Healthcare, Consulting, Telecom.

Skills you need to become a Data Architect

SkillImportance
Data Modeling10/10
Data Warehousing10/10
Cloud Data Platforms10/10
SQL9/10
Data Governance9/10
Lakehouse Architecture9/10
ETL/ELT Design9/10
Big Data (Spark/Hadoop)8/10
Data Security & Compliance8/10
Stakeholder & Solutioning8/10

Core tools: Snowflake / Databricks, ERwin / dbdiagram, Apache Spark, Collibra / Purview, Azure / AWS / GCP, dbt.

Data Architect learning roadmap

Beginner · 3-5 months

Foundations & core tooling

Build: Design a normalized + star-schema data model for a sample domain.

Intermediate · 5-6 months

Applied, real-world builds

Build: Architect a cloud warehouse/lakehouse with governance and ETL design.

Advanced · 6-8 months

Production, scale & specialization

Build: Deliver an enterprise data architecture with lakehouse, governance and compliance controls.

Get a day-by-day Data Architect study plan →

10 Data Architect portfolio projects

Normalized + Star Model

Beginner

Design OLTP and star-schema models.

Skills: Data Modeling, SQL

Warehouse Design

Beginner

Design a warehouse for a domain.

Skills: Data Warehousing, Data Modeling

Lakehouse Architecture

Intermediate

Bronze/silver/gold lakehouse design.

Skills: Lakehouse, Data Modeling, Cloud

Governance Framework

Intermediate

Catalog, ownership and access design.

Skills: Data Governance, Data Modeling

ETL/ELT Design

Intermediate

Design ingestion and transformation.

Skills: ETL/ELT Design, Data Modeling

Cost & Performance Plan

Intermediate

Partition/cluster and cost strategy.

Skills: Data Warehousing, Cloud

Enterprise Data Platform

Advanced

Full platform: ingest -> govern -> consume.

Skills: System Design, Lakehouse, Governance

Master Data Management

Advanced

MDM design for a key entity.

Skills: Data Modeling, Data Governance

Migration Architecture

Advanced

Legacy-to-cloud data migration design.

Skills: Cloud, Data Modeling, ETL

Data Mesh Design

Advanced

Domain-oriented data product design.

Skills: Data Modeling, Governance

How to build three of these Data Architect projects

Each brief names the skills from the list above that the project proves, and what to be ready to explain about it.

Normalized + Star Model

Take one business process you know, such as food delivery orders, and model it twice: a normalized OLTP schema and a star schema with one fact table. Write the fact table's grain in one sentence before adding any columns.

  • Data Modeling
  • SQL
  • Data Warehousing

Be ready to explain: Why each dimension sits where it does, and one query that is simpler on the star schema.

Lakehouse Architecture

Lay out bronze, silver and gold layers on Databricks or Snowflake for the same orders data, with the Spark or SQL jobs that move data between the layers.

  • Lakehouse Architecture
  • Cloud Data Platforms
  • Big Data (Spark/Hadoop)

Be ready to explain: Where deduplication happens, and how you would replay one bad day of data.

Governance Framework

A one-page plan for the same data: an owner for each table, which columns hold personal data such as phone numbers, who may read them, and how they are masked.

  • Data Governance
  • Data Security & Compliance
  • Stakeholder & Solutioning

Be ready to explain: One access decision you would defend to a business team that wants more data than it needs.

Common Data Architect interview questions

What is a slowly changing dimension? Types?Hard

What they're testing: Track history; Type 1 overwrite, Type 2 versioned

ETL vs ELT — when to use each.Medium

What they're testing: Transform before vs in the warehouse

How do you manage and optimize cloud cost?Medium

What they're testing: Right-size, autoscale, reserved, monitoring

What is an index and what are its trade-offs?Medium

What they're testing: Speeds reads, slows writes, uses storage; B-tree on filter/join cols

How do you handle schema evolution in pipelines?Hard

What they're testing: Contracts, versioning, backward compatibility

When to use broadcast join vs shuffle join.Medium

What they're testing: Small table broadcast avoids shuffle

Fact vs dimension tables.Easy

What they're testing: Measures/events vs descriptive context

How do you ensure data quality in a warehouse?Medium

What they're testing: Tests, constraints, monitoring, contracts

Explain VPC/networking basics in the cloud.Medium

What they're testing: Subnets, routing, security groups

Explain a CTE vs a subquery vs a temp table.Medium

What they're testing: Readability, reuse, materialization and scope differences

Where do you put data-quality checks in a pipeline?Medium

What they're testing: At ingestion and post-transform with alerting

How does partitioning affect performance?Medium

What they're testing: Parallelism and shuffle volume

Practice the full Data Architect question bank →

More Data Architect interview questions, answered

What is the grain of a fact table and why must it be declared first?

The grain is exactly what one row of the fact table means, such as one row per order line. Declare it before adding measures or dimensions, because every column must be true at that grain; mixing grains is how totals end up double-counted.

More in Data modeling interview questions →

What are conformed dimensions and why do they matter?

A conformed dimension, such as customer, product or date, is one shared version used by several fact tables, so revenue by customer and tickets by customer agree. Without it, each team builds its own customer table and the dashboards stop matching.

More in Data modeling interview questions →

Compare the Kimball and Inmon approaches.

Inmon builds a normalized enterprise warehouse first and feeds data marts from it. Kimball builds star-schema marts one business process at a time and joins them through conformed dimensions. Most modern stacks mix the two: a cleaned integration layer feeding Kimball-style marts.

More in Data modeling interview questions →

You're asked to build a warehouse from scratch for a company that has none. Walk through your approach.

Start from the decisions the business cannot make today, pick one high-value process, and plan the shared dimensions on a bus matrix before writing code. Then profile the sources, choose the platform and layers, deliver that one process end to end, and repeat.

More in Data warehouse interview questions →

Why is separation of storage and compute considered the defining feature of cloud warehouses?

Because each can scale on its own. Years of history no longer force a bigger cluster, ELT jobs and dashboards can run on separate compute against the same data, and compute can pause when idle, so you pay for queries rather than for a cluster sized for the peak.

More in Data warehouse interview questions →

Certifications for Data Architects

  • AWS Certified Solutions Architect - AssociateAmazon Web Services · Very High value
  • Google Cloud Professional Data EngineerGoogle Cloud · Very High value
  • SnowPro Core CertificationSnowflake · Very High value
  • Microsoft Certified: Azure Solutions Architect Expert (AZ-305)Microsoft · Very High value

Data Architect career path

Data Architect -> Senior Data Architect -> Chief Data Architect

Related roles: Solution Architect, Data Engineer, Cloud Architect

Frequently asked questions

What skills do you need to become a Data Architect?

Core skills include Data Modeling, Data Warehousing, Cloud Data Platforms, SQL, Data Governance. Show a clear architecture with governance and trade-offs.

What projects should a Data Architect build for a portfolio?

Strong starter projects: Normalized + Star Model; Warehouse Design; Lakehouse Architecture; Governance Framework.

How long does it take to become job-ready as a Data Architect?

A focused plan runs roughly 3-5 months for fundamentals, then applied projects. Difficulty rating: 8/10.

What is the career path for a Data Architect?

Data Architect -> Senior Data Architect -> Chief Data Architect

Ready to become a Data Architect?

PrepNPlaced turns this guide into a plan: a day-by-day roadmap, an ATS-ready resume and interview practice.

Start free →