Head of Data · Analytics Engineering · Data Platforms

Juan Caravaca

Head of Data and technical data leader building AI-ready analytics organizations.

I build and lead multidisciplinary Data teams while staying close to the architecture — dbt and Analytics Engineering, semantic models and governed metrics, modern data platforms, and analytics that AI agents can actually be trusted to use.

Head of Data at Northius · 13+ years across IBM, visualNAcert, CoverWallet (Aon) and Northius

Portrait of Juan Caravaca

In numbers

13+
years in Data
and technology
10
people in the Data team,
five disciplines
14
brands supported
across four markets
500+
dbt models
in production
01 —

Leadership without leaving the technology behind

I lead the Data function the way I think it should be led: close enough to the architecture to make real technical decisions, and far enough back to design the team, the operating model and the priorities that turn data work into business impact. In practice that means I stay directly involved in architectural decisions, still write dbt models and SQL, review data models and semantic definitions, and mentor the team — alongside hiring, career development and making sure Data improves how the business decides.

Leadership

  • Data strategy and operating model — how the function is organized, prioritized and held accountable
  • Team leadership across five disciplines: Data Engineering, Analytics Engineering, Analytics, BI and Data Science / Product Science
  • Hiring and career development — building progression paths in a multidisciplinary team
  • Executive stakeholder alignment — translating business strategy into data priorities, and data reality back into decisions
  • Governance — metric ownership, definitions and data quality as shared organizational assets
  • Cross-functional partnership with Marketing, Sales, Finance, Product and academic operations

Technical depth

  • Analytics Engineering with dbt — modeling, tests, exposures, lineage and software-engineering practice applied to analytics
  • Semantic modeling and governed metrics — one set of definitions for humans, BI tools and AI agents
  • Modern data platforms — AWS, Iceberg, Athena/Trino, Snowflake and Redshift across different environments; orchestration with Prefect and Airflow
  • Data modeling and architecture — SQL and Python, hands-on
  • BI architecture and BI-as-Code — applying version control, reusable models and software-engineering practices to analytics delivery
  • Agentic analytics — governed, semantically described data access for AI agents over MCP
02 —

The Data-to-Decision Chain

Most companies don't have a data problem. They have a decision problem: data everywhere, and still no reliable way to turn it into action — and now AI stacked on top of the same broken foundation. This is the chain I build. Build it whole and in order, and the organization decides well. Miss a link, and everything below it wobbles.

Strategy
What the business is trying to win, stated clearly enough to build on.
Fails whenevery team optimizes its own patch and nobody knows what winning means.
Metrics
The few numbers that define winning, governed and shared.
Fails whenevery department calculates the same KPI differently.
Semantic models
One language for the business, so every question means the same thing.
Fails whenevery query is a negotiation over what "customer" means.
Data platform
Reliable, so people and AI agents alike can build on it.
Fails whennobody trusts the number, so decisions default to gut.
AI
Applied where it changes the decision, not where it looks good in a demo.
Fails whenpilots never reach production.
Impact
Decisions made, actions taken, results that show up in the business.
Fails whenthere are lots of reports and few decisions.

Technology enables it. People and the operating model create it.

Read the essay behind this →
03 —

Experience

Engineering and Big Data, then technical leadership as CTO, then Analytics Engineering and data platforms, now Head of Data. The same thread throughout: data systems that hold up when a real decision depends on them.

Northius
Oct 2024 – Present
Private-equity-backed education group — 14 brands across Spain, Portugal, the UK and LATAM.
Head of Data
  • Lead the Data function for the group, supporting 14 brands across four markets.
  • Lead a 10-person multidisciplinary team spanning Data Engineering, Analytics Engineering, Analytics, BI and Data Science / Product Science — including hiring, career development and the operating model.
  • Own the platform architecture and stay hands-on in it: AWS, Apache Iceberg on S3, Athena/Trino, Prefect, dbt and Superset — 500+ dbt models and 40 source systems under a 3h freshness SLA.
  • Built governed metrics and semantic modeling into the platform — 300+ governed metrics serving 300+ internal BI users through a production semantic layer.
  • Working on agentic analytics: giving AI agents governed, semantically described access to enterprise data over MCP, instead of relying on direct, ungoverned warehouse access.
  • Introduced Analytics Engineering and BI-as-Code, turning analytics and BI into reusable data products the team can evolve without breaking what is already in use.
  • Established a cross-functional product-analytics practice, shifting Data from a reactive reporting service to co-creation with Marketing, Sales, Finance, Product and academic operations.
CoverWallet, an Aon company
Sept 2021 – Oct 2024
Insurtech acquired by Aon (NYSE: AON).
Analytics Engineer → Data Platform Manager
  • As Data Platform Manager (2023 – Oct 2024), led a team of 7 — four Data Engineers and three Analytics Engineers — covering Data Engineering and Analytics Engineering.
  • Led the dbt adoption project, moving analytics onto a modular, layered data model that could scale with the business instead of accumulating one-off SQL.
  • Rebuilt the Looker implementation on top of a governed, universal dbt data model — reducing coupling and improving scalability.
  • Built and modeled the analytics platform on Redshift, then Snowflake, orchestrated with Airflow on Kubernetes; SQL and Python day to day.
  • Joined post-acquisition and contributed to integrating CoverWallet's data platform and reporting into Aon.
visualNAcert
2015 – 2021
Agritech — data products for agronomic operations.
Big Data Analyst → CTO
  • Led technology and product engineering as CTO (2018–2021), building data-driven products used in agronomic field operations.
IBM Software Group
2013 – 2015
Big Data Technical Specialist
  • Worked on data platforms and distributed data technologies during the early enterprise adoption of Big Data.
  • Selected for IBM's Spark 40 in 2015, a European specialist cohort focused on Apache Spark during its early enterprise adoption.
Earlier Machine Learning researcher, iDAL — Intelligent Data Analysis Lab, University of Valencia (2011–2013).
05 —

Education & training

Education

University of Valencia
MSc, Electronics Engineering
2012
University of Valencia
Engineer's degree, Electronics Engineering
2009–2011
University of Valencia
BTech, Telecommunications Technology
2005–2009

Programs & technical training

LIDR.co — Ignite Program
Technical leadership
2023
Anthropic
Agent Skills · Model Context Protocol (MCP) · Building with the Claude API · Claude Code in Action
2026

Let's connect.

Happy to talk about data organizations, Analytics Engineering, or what it actually takes to make analytics reliable enough for AI agents.