Data Scientist

Il y a 1 semaine

Antwerpen, Antwerpen, Belgique Jobtailor Temps plein 60 000 € - 90 000 € Contrat

Gratuit avec email ou Google

Enregistrez cette offre et organisez votre recherche

Créez un compte gratuit pour enregistrer des offres d'emploi, créer des alertes et revenir à cette liste depuis votre tableau de bord.

Gratuit avec email ou Google

As our Applied Data Scientist, your mission is to build the machine‑learning models that make Yuki smart — the models that classify documents and predict the right general ledger (grootboek), turning a client's raw invoices into clean, usable accounting data.
This is a modeling role with real production discipline. You won't just train models and walk away; you'll keep them accurate as the data and the business shift.
You’ll co‑create with engineers and product people, and you’ll grow a genuine understanding of how accountants actually think about a document.
Build the models: Train and evaluate classic ML (XGBoost, scikit‑learn) and transformer models for document classification and ledger prediction.
Fine‑tune LLMs: Do real model training and fine‑tuning with PyTorch and HuggingFace.
Design experiments: Own feature engineering, model evaluation, and selection, plus embedding and retrieval strategy for vector search.
Guard production quality: Run drift detection, eval pipelines, and the MLflow model registry so models hold up over time.
Learn the domain: Build deep understanding of accounting and financial‑document structure, and translate it back into modeling decisions.

A strong applied ML background, comfortable across both classic ML and transformer models.
Hands‑on with the stack: Python, PyTorch, scikit‑learn, HuggingFace, Pandas, MLflow, Databricks, Snowflake.
Rigorous about evaluation, you ship models that survive contact with production, not just the notebook.
A communicator and co‑creator: strong IC who thinks, shares, and explains ML in plain business language (data‑driven storytelling and visualization a real plus).
Nice to have: financial/accounting domain exposure and experience with large document corpora.
Fluent English is a must.