Senior MLOps for AI Product

📋 Description

• • Own the end-to-end lifecycle of machine-learning infrastructure that powers Qonto’s AI-driven finance workspace, serving 600,000+ European SMEs and freelancers. You will architect, build, and continuously improve real-time and batch pipelines that move billions of financial events through training, validation, and low-latency inference stages every day.
• • Design resilient, observable Python micro-services that package, version, and deploy models in Docker/Kubernetes clusters across multiple regions (Paris, Berlin, Milan, Barcelona, Belgrade). You will embed CI/CD, automated testing, canary releases, and rollback strategies so that new models reach production safely within hours, not weeks.
• • Optimize data storage and access patterns on Snowflake and S3 to balance cost, latency, and throughput. You will choose partitioning schemes, indexing strategies, and caching layers that let ML Engineers iterate on feature sets without exploding compute budgets.
• • Implement Kafka-based stream processing that turns raw transaction streams into real-time features for fraud detection, cash-flow forecasting, and automated bookkeeping suggestions. You will guarantee exactly-once semantics, back-pressure handling, and sub-second end-to-end latency.
• • Deploy and manage model-serving systems (TensorFlow Serving, FastAPI, or BentoML) behind load-balanced endpoints, ensuring 99.9 % availability, autoscaling, and A/B testing capabilities. You will set up Prometheus/Grafana dashboards and PagerDuty alerts so anomalies are caught before customers notice.
• • Partner with ML Engineers, Backend Engineers, and Product Managers to translate business KPIs into technical SLAs. You will run weekly architecture reviews, post-mortems, and capacity-planning sessions to keep the platform two steps ahead of customer growth.
• • Champion data-quality initiatives: build automated schema validation, anomaly detection, and drift monitoring that feed back into retraining schedules. You will make “bad data” visible and actionable within minutes, not days.
• • Contribute to Qonto’s open-source culture by publishing internal libraries, writing technical blog posts, and presenting at meetups. You will mentor junior engineers and run internal guilds on MLOps best practices.
• • Embrace lean methodology—ship small, measure impact, iterate fast. You will instrument every pipeline with metrics that tie infrastructure changes to customer outcomes like reduced manual bookkeeping time or faster invoice reconciliation.
• • Work hybrid from one of our five European hubs, collaborating in person at least two days per week. You will join a 55 % international, 44 % women, and 20 % parents workforce that values psychological safety and continuous learning.