About jovuloe

Where patterns
in data become
practical skill

jovuloe is a workshop platform built around unsupervised learning in finance - the kind of work that reveals structure in data before you know what you're looking for.

6+ Years running
live workshops

38 Countries
represented

14 Active workshop
programmes

jovuloe workshop session with participants working through financial data

Our background

Started with a specific problem

In 2018, a group of analysts and educators noticed the same gap across financial institutions: teams could run supervised models well, but struggled when the data arrived without clear labels or defined outcomes. Clustering, dimensionality reduction, anomaly detection - these techniques existed, but the training for them was either too theoretical or too generic to be useful in finance.

jovuloe was built to close that gap. The platform runs structured workshops where participants work through real financial datasets - portfolio groupings, market regime detection, transaction pattern analysis - using tools like scikit-learn, UMAP, and HDBSCAN in actual code environments, not slide decks.

Each workshop is designed around a concrete task, not a curriculum. You finish with something you built and can explain, not just a certificate.

What the platform covers

Clustering methods applied to asset returns and transaction data. Dimensionality reduction for risk factor visualisation. Anomaly detection in financial time series. Practical model evaluation without ground truth labels.

Clustering

Anomaly
detection

Visualisation

The people behind it

Instructors & leads

Each instructor has worked in financial data roles before moving into education. They teach what they have actually used.

Orlaith Devane

Lead Instructor - Unsupervised Methods

Orlaith spent eight years working on quantitative risk at two European asset managers before joining jovuloe. She designs most of the clustering and regime-detection workshops and insists on using messy, real datasets rather than cleaned examples.

Benedikt Szábo

Data Engineering Lead

Benedikt handles the technical infrastructure behind the workshop environments and contributes to anomaly detection modules. He has a background in financial software engineering and cares a lot about reproducibility.

Fionnuala Ó Briain

Curriculum & Research

Fionnuala researches how practitioners actually learn technical methods and uses that to shape the workshop structure. She previously taught applied statistics and has a particular focus on evaluation methods when labels are absent.

How we work

The structure behind each workshop

Every programme follows a logic that puts the dataset first and the technique second - because that is how the work actually arrives.

Data before theory

Each session opens with a dataset and a question, not a lecture. Participants explore the data first, then reach for the appropriate method. This order matters - it keeps the technique grounded in something concrete rather than abstract.

Collaborative exercises

Workshops run in small groups where participants share observations and challenge each other's interpretations. Financial data rarely has a single correct reading, and working through disagreements in a structured setting builds the kind of judgement that solo study cannot.

Honest evaluation

Unsupervised methods are hard to evaluate. We spend deliberate time on this - silhouette scores, stability testing, domain sense-checking - because a model that cannot be evaluated cannot be trusted in production.

Accessible from anywhere

Sessions run across time zones with asynchronous materials available for participants who cannot attend live. The platform is built for a distributed audience - most participants are working full-time and fit workshops around existing responsibilities.

jovuloe collaborative workshop environment

"The goal is not to produce people who can run a clustering algorithm. It is to produce people who know when not to."

Orlaith Devane - Lead Instructor

Where patternsin data becomepractical skill