QuantAIProductLeadership
Management Consultant · Lead AI Engineer
Scroll
Management Consultant · Lead AI Engineer
I take the rigour of econometrics and quantitative finance and turn it into AI products that people in risk, compliance and finance actually use: from the first client conversation to architecture, cloud deployment, training and rollout.
Today I lead AI engineering at Assets Only, a consultancy for risk, compliance and ESG in the Netherlands. Before that I modelled interest rates on GPUs for insurers and pension funds.
Quant
AI Engineering
Product
Rigour into results.
What I bring
Most people are a quant, an engineer, a consultant or a product lead. The interesting work happens where those meet. Open one to see what's inside.
LLMs in production, with people in control.
Econometrics, stochastic models, scenario engines.
Knowing the rules well enough to build for them.
From a client's problem to a product they adopt.
Selected work
Products, models and systems from the last few years. Most were built at Assets Only together with colleagues and clients.
09projects
01 / 09 · Assets Only
A workbench for financial crime teams
Customer due diligence from onboarding to decision: document handling, configurable risk assessment, identity verification, PEP and sanctions screening, four-eyes review and a full audit trail. AI drafts English and Dutch case text from dossier facts. The decisions stay with the analyst.
clientinsight.nl ↗02 / 09 · Assets Only
Healthcare integrity for regions and municipalities
One workspace that brings together open sources (annual accounts, supervisory reports), risk indicators and a direct channel to care providers. I led the development team and the move to production on Azure, with single sign-on, enforced two-factor authentication and EU-hosted AI.
zorglens.nl ↗03 / 09 · Assets Only
Enterprise risk management, built from scratch
A risk and control platform aligned with COSO and ISO 31000: risk and process taxonomies, risk-control tables with gross and net scoring, control testing, remediation tracking and four-eyes approval. Multi-tenant with row-level security and MFA, now growing modules for ISO 27001, DORA and NIS2.
04 / 09 · Assets Only
Greenwashing risk, scanned
Scans company websites against the new EU rules on green claims, flags the claims most likely to cause trouble and explains why. Clients get an action dashboard and monthly monitoring. Several AI models check each other's findings before a person signs off.
05 / 09 · Assets Only
Getting a whole consultancy to use AI, safely
Tool selection, role-by-role efficiency analyses and a business case for the partners. A three-tier AI usage framework, a processing register that includes every AI tool, vendor agreements and DPIA planning. Plus guides and training so non-technical colleagues actually pick it up, and workshops on AI for one of Europe's largest pension investors.
06 / 09 · Assets Only
26 training films, six courses, quizzes
An in-app academy that trains colleagues to present and run the products, with a quiz per course and a team dashboard. For ZorgLens I made explainer films and a LinkedIn teaser series, scored with original music from a small engine I wrote that plays sampled instruments.
07 / 09 · MavenBlue
Stochastic interest-rate modelling in CUDA
Implemented a stochastic-volatility interest-rate model with displaced diffusion in GPU-accelerated C (CUDA) and Python, and reconciled it against the reference implementation. Built a risk-neutral economic scenario generator, wrote the mathematical documentation and supported insurers and pension funds on ALM and ORSA studies.
08 / 09 · Assets Only
Six analyses for sustainable investors
SFDR adverse-impact modelling with Monte Carlo classification, an NGFS climate stress engine with carbon-price paths and NAV fan charts, carbon-adjusted valuation and attribution, ESG rating reconciliation with PCA, and biodiversity factor pricing with Fama-MacBeth regressions. Built on stylised data calibrated to public benchmarks.
09 / 09 · Own work
I design the system, agents do the typing, I review
An always-on server where several AI coding agents build, review and test in parallel, and a brief is waiting every morning. One of my own projects went through eight rounds of a multi-agent bug hunt: 190 bugs found and fixed, and the test suite grew from 117 to over 700 tests.
Path
Willis Towers Watson
University of Amsterdam · then VC Analyst, portfolio management at Cydonia Fund
QuantFi, during my master's
University of Amsterdam · thesis on cointegrated cryptocurrency prices and statistical arbitrage
MavenBlue · interest-rate models, scenario generators, ALM and ORSA
Assets Only · Lead AI Engineer since April 2026, Management Consultant since June 2026
Building and leading teams where quantitative rigour and AI come together, in finance and beyond.
Off the clock
On the side I publish small apps and games as LukasOnAir →