Ian Ronk
Ian Ronk
Data & AI Architect · Project Lead
◎ Amsterdam, NL

Data and AI systems
that run
in production.

I design and build data and AI systems end to end, and lead the projects that deliver them.

§ 01About

Hi, I'm Ian.
I build systems that keep running.

I am Ian Ronk, a Data & AI Architect based in Amsterdam. I build systems end to end, from the servers and pipelines underneath to the models and workflows people actually use, and I lead the projects that get them into production.

I spent five years at KR&A, the first four part-time alongside my studies, growing from junior data scientist to Head of Data / Data & AI Lead. There I was technical lead on a 13-country house-price index study for Eurostat, whose results Eurostat published in a statistical working paper, and led the Connectivity Score, an accessibility score for 38 countries that a.s.r. real assets now uses in its fund-analysis platform. I also build AI and automation systems on Azure and on infrastructure I set up and run myself.

More about me
Ian Ronk
Since 2019
building software that runs in production
5 years
of production data systems at KR&A
38
countries covered by the live Connectivity Score
300k
records a week through one scraping pipeline
4
engineers in the data team I built up
§ 02Selected work

The proof behind my ideas.

From a published Eurostat study to systems in daily use.

Illustration
  • Walk
  • Drive
  • Transit
Places in reach
  • Walk7
  • Drive40
  • Transit13
Score
  • Walk11
  • Drive63
  • Transit20
GET /score?address=example
200 OK
{
  "walk": 11,
  "drive": 63,
  "transit": 20
}

Streets and transit timetables become one network per region.

§ 02.01

Connectivity Score: accessibility for 38 countries

The problem
Real-estate investors needed to compare how reachable a location is, on foot, by car and by public transport, across continents. Existing scores mostly cover one region and walking only.
What I built

As project lead at KR&A: a method grounded in academic accessibility research, and the system that computes it: one multi-modal graph per region, a custom routing engine that replaced pgRouting, and a 13-server distributed build over terabytes of data. The result is an API that scores any address in 38 countries, with every score decomposable into the counts and distances behind it, documented in a methodology paper.

Outcome
Live as an API; a.s.r. real assets uses it in its PULSE fund-analysis platform (announced September 2026).
CompetencesData AnalysisSystem and Data InfrastructurePythonPostGISRoutingDistributed computingS3LinuxFastAPI
Read the full case
§ 02.02

Modular Design: sales, CRM and internal tools

The problem
In many companies the same internal processes are repeated by hand: collect what is needed, check it against the rules, decide, and keep track as things change. It is slow, hard to audit and easy to get wrong.
What I built

Systems I architect and build on one principle: agents propose, people approve. Each system assembles its workflows from small blocks, so a new workflow starts from what already works in that system. For example:

  • An order-intake agent and light ERP for a trading company. Orders, quote requests and negotiations arrive by email in three languages; anything that needs a decision goes in a "needs you" queue.
  • An internal CRM and outreach engine for a small B2B data firm. Every AI-written email is saved as a draft; a person sends it.
  • A modular internal platform. A person decides, and every decision can be traced.
Outcome
The order system is in acceptance testing with the owner after four test rounds, with about 2,000 automated tests; the CRM runs internally; the platform is in progress.
CompetencesAISystem and Data InfrastructureAI agentsWorkflow engineModular architectureEmail automationCRM
Read the full case
log(price) = β · features + δ(month)
  • Monthly index from offers
  • Official quarterly index
§ 02.03

Eurostat house-price index: 13 EU countries

The problem
Official EU house-price indices are quarterly and lag; Eurostat wanted to know whether web-scraped asking prices can produce sound monthly indices.
What I built

As technical lead at KR&A, for eight people across two organisations: a pipeline on Azure scraping property portals in 13 countries (30 million usable offers on 6.7 million properties), cleaning, matching and geo-enrichment, and the index system on top: hedonic regressions with typed per-source configuration, so a new country's index takes hours instead of days.

Outcome
Eurostat published the results in a statistical working paper (2026).
CompetencesData AnalysisSystem and Data InfrastructurePythonScrapy/SeleniumAirflowIcebergMongoDBPostGISRAzure
Read the full case
LanguageBuddy home screen with a demo profile: level per skill, words due, speaking band, today's conversation and the daily practice checklist
LanguageBuddy logo
§ 02.04

LanguageBuddy: AI language tutor

The problem
Language apps drill words out of context; I wanted a tutor that adapts to what I actually get wrong.
What I built

A self-hosted tutor for Dutch, Italian and Spanish (A1 to C1): chat and voice calls with an LLM tutor, a CEFR placement exam, and a spaced-repetition queue that turns every mistake into the next day's exercises, fed by a real-news scraper for reading practice.

Outcome
In daily use; 6,200+ vocabulary entries and 177 conversation scenarios.
CompetencesAIFastAPILLMSpeech (TTS)SQLiteDockerLinux
Read the full case
More projectsSponsoredBye, FishFinder, and research on geodata and spatial data.All projects
§ 03Credentials

Three areas, shown in the work above.

What I work with, and where to see it: each area links to the case studies and projects that show it.

§ 04Experience

Work and education.

Five years at KR&A, the first four part-time alongside my studies, from junior data scientist to Head of Data / Data & AI Lead, and the systems I build now. Each role links to the projects behind it.

  1. 2026 – present

    Independent work

    AI and automation systems: an order-intake agent, a CRM and a modular workflow platform.

  2. Jul 2025 – Oct 2026

    Head of Data / Data & AI Lead

    KR&A · Amsterdam

    Data and AI architecture and delivery: a data team of four, eight production servers, the Connectivity Score from method to live API, and LLM/OCR document extraction.

  3. Jun 2022 – Jul 2025

    Data Scientist & Technical Lead

    KR&A · part-time

    Technical lead of the Eurostat house-price index study (eight people, two organisations); moved the data platform to Airflow, Iceberg and FastAPI.

  4. Oct 2021 – Jun 2022

    Junior Data Scientist

    KR&A · part-time

    Flood-risk prediction from local features, and a spatial API made ten times faster.

    • Sep 2019 – Oct 2021

      Junior Full-Stack Developer

      Exact (former SRXP) · part-time

      Enterprise expense software under CI/CD and testing.

      Education

      1. 2023 – 2025

        MSc Data Science & Business Analytics

        Bocconi University

        Focus on finance, econometrics, statistics and NLP.

      2. 2019 – 2023

        BSc Artificial Intelligence

        University of Amsterdam

        Minor in Linguistics, and an Erasmus minor at the University of Bologna.

      Download CV (PDF)Full CV
      § 05Publications & writing

      Publications & writing.

      Publications

      All publications →

      Writing

      All writing →
      § 06Contact

      Working on a data or AI system? Let's talk.

      Get in touch

      Twenty minutes to talk it through, or just send a message.