Staff Software Engineer (Machine Learning)

Join our team as a Staff Software Engineer (Machine Learning) and take ownership of the technical direction for our ML stack. You will be responsible for feature definition and computation, training data generation, training infrastructure, model serving, and production monitoring. You will also build tooling for training/serving consistency, design model linkage to datasets, and enable data scientists to generate reproducible datasets. Additionally, you will measure delivery time, engineering effort, and rework, and use that evidence to prioritize improvements. In your first year, you will establish a delivery baseline, extend capabilities to additional models, standardize model monitoring, evaluate build-versus-buy options for ML platform tooling, and partner with engineers, data scientists, and analysts.

Missions

• Assumer la direction technique de l'ensemble de la pile ML, de la définition des fonctionnalités à la surveillance de la production.
• Construire des outils pour garantir la cohérence entre l'entraînement et le service, en tenant compte des différences de sources de données.
• Concevoir comment chaque modèle déployé reste lié à son ensemble de données, aux versions des fonctionnalités, aux étiquettes et au code d'entraînement.

Profil recherché

- 8+ years of overall software engineering experience, including strong production skills in Python and SQL, experience building production systems in a JVM language, and 3+ years of experience building and maintaining ML platforms

- Feature store, feature platform, or ML platform experience at a company where models make consequential decisions is a plus

- Sound reasoning about time in data: point-in-time correctness, label leakage, feature availability, and training/serving skew

- Strong communication skills, with the ability to explain trade-offs clearly and find workable solutions across engineering, data science, risk, marketing, and finance

- Experience with streaming and change data capture, large-scale batch on Apache Beam or Spark, or distributed training is a plus

- Experience setting a long-term technical direction and turning it into an achievable roadmap, delivering useful improvements along the way

- A track record of improving ML delivery workflows, and the ability to explain the trade-offs, results, and lessons from those decisions

- Experience in financial services, credit, fraud, or another regulated decisioning domain, and familiarity with model risk management, is a plus

- Experience leading initiatives from an ambiguous problem through scoping, stakeholder agreement, and delivery

- 3+ years experience building and operating ML systems in production, including feature pipelines, the training data path, the serving layer, and the monitoring around them

- Fluency with AI tools, including coding agents, in your own engineering work, with the judgment to evaluate their output and own the quality of what you ship