Writing
Industry perspectives on process ownership and data quality, alongside the applied publications and research that show the technical foundation behind my work.
Reliable AI agents depend on business-process context and framed autonomy: shared case identifiers, defined process states, and clear boundaries on what an agent may change.
Read at AI FinlandAI work creates value only when process ownership and the resulting action are clear.
Read at Vuono GroupFive recurring process problems that produce unreliable data, with questions for finding the cause.
Read at Vuono GroupThis practical guide covers quality assurance across the full environmental water-monitoring lifecycle, from observation and storage to validation, reporting, and reuse.
Read the public guideA co-authored method for finding outliers in river and groundwater monitoring series by using the correlation between related measurement points, published in Hydrology Research.
Read the paperA comparison of methods for learning from highly imbalanced ecological image data, with practical guidance for identifying rare taxa.
A first-author paper from the Amazon collaboration on learning from rankings or comparisons and evaluating several candidates in parallel.
A first-author collaboration with an Amazon researcher on making Bayesian optimisation more efficient by encoding prior knowledge about where the best solution is unlikely to be.
A first-author study with Novartis on modelling organ maturation and evaluating pediatric drug-dosing regimens with limited data.
My doctoral dissertation looks at how Gaussian processes can learn from comparisons, rankings, and other feedback when exact measurements are costly or unavailable.
Practical advice for optimisation problems with many structured inputs and expensive evaluations.
