Turn Your Data Into Decisions That Scale.
ML Ventures helps growth-stage companies build AI-powered analytics systems, predictive models, and data infrastructure — without the overhead of a full team.
The Person Behind the Work
I'm Michael, a mathematician and engineer who found his way into data. I optimized aircraft maintenance scheduling at Southwest Airlines, worked in actuarial analytics at John Hancock, and led revenue analytics at HousingAnywhere before launching ML Ventures.
I hold a BS in Mathematics and Aerospace Engineering and an MS in Econometrics: Quantitative Finance from Erasmus. I've studied across 5 countries and traveled to 65+. SQL clicked for me at Southwest and became the foundation of everything I build.
Today I help small businesses turn messy data into clear decisions, through dashboards that get used, forecasting models grounded in real business context, and pipelines that save hours every week. I combine technical depth with business thinking, and I'm fully AI-native in how I work.
3 degrees
Including an MSc in Econometrics from Erasmus University
65+ countries
Studied across 5, traveled extensively across the rest
4 corporate jobs
Southwest Airlines, John Hancock, HousingAnywhere, Vintus
AI-native
Claude Code in daily development workflow
4 languages
English, Dutch, Mandarin and Japanese
10+ core tools
Python, SQL, dbt, Airflow, Snowflake, Azure, scikit-learn, OpenAI and more
3 service pillars
Dashboards, forecasting and data pipelines
3 continents
Based seasonally across the Netherlands, China and the USA
What I Build
From raw data to production-ready systems. I work across the full analytics stack.
AI Workflow Integration
Connect your existing tools to AI models that actually work. I design and build AI-assisted pipelines — from LLM-powered reporting to automated insight generation — that fit your current stack.
Data & BI Systems
End-to-end analytics infrastructure: data warehouses, dashboards, and reporting pipelines built for decision-makers. Clean, maintainable, and built to grow with you.
Predictive Analytics & ML
Forecasting, classification, and anomaly detection models built from your data. From churn prediction to demand forecasting — deployed and maintained, not just prototyped.
Data Engineering
Pipelines that move, transform, and deliver data reliably. dbt, Airflow, cloud-native ETL — infrastructure you can trust at scale.
Selected Work
Details anonymized per client agreements.
A premium Windows desktop launcher built for developers. One click opens a full 2×2 Windows Terminal layout, File Explorer, and any startup commands — all configured per project via a dark, theme-able GUI.
- →Instant project environment launch from a single click
- →Per-pane terminal paths and commands, drag-and-drop category management
- →Packaged as a standalone .exe with startup integration
Led a full reconciliation of a platform's revenue data across multiple sources, identifying and resolving discrepancies that had gone undetected for months.
- →Traced and resolved revenue mismatches across billing, payments, and bookings
- →Built reconciliation framework used as the ongoing source of truth
- →Enabled accurate financial reporting for leadership and investors
Designed and built automated daily dashboards for executive and finance teams, replacing manual reporting with real-time visibility into core business metrics.
- →Automated daily refresh of KPIs across revenue, growth, and operations
- →Adopted company-wide as the primary decision-making tool
- →Reduced manual reporting effort by several hours per week
Repaired and stabilized a set of broken data pipelines that were causing inconsistencies across analytics and reporting systems.
- →Identified and fixed root causes of data quality issues
- →Rebuilt pipeline logic for reliability and maintainability
- →Established monitoring and alerting for pipeline health
End-to-end data platform build on cloud infrastructure, from raw ingestion through transformation to analytics-ready outputs for business stakeholders.
- →Architected and deployed a modern data stack from scratch
- →Integrated multiple source systems into a unified data model
- →Enabled self-serve analytics for non-technical teams
Developed machine learning models to predict key business outcomes, delivering actionable insights that directly informed operational decisions.
- →Built and validated predictive models using real business data
- →Delivered predictions integrated into existing workflows and tools
- →Improved decision accuracy for targeted business processes
Built custom web apps and Excel-based tools to give teams fast, reliable access to data without needing technical expertise.
- →Delivered interactive web apps for data exploration and reporting
- →Created Excel-based tools for teams more comfortable in spreadsheets
- →Reduced dependency on ad-hoc analyst requests
Let's Talk About Your Data.
No sales pitch. Just a 30-minute conversation about where you are, where you want to be, and whether I'm the right fit to help.
Book a Free 30-Min Chat →Michael Castelein
LinkedIn