Machine Learning Engineer

• 5 to 8 years of experience in ML engineering, MLOps or backend infrastructure with ML systems in production.
• Strong software engineering fundamentals; comfortable owning services end to end.
• Experience with model serving frameworks (Seldon, KServe, BentoML, TorchServe or similar) and orchestration tools (Airflow, Kubeflow, MLflow or similar).
• Hands on experience with streaming systems (Kafka, Kinesis, Flink or similar).
• Familiarity with containerization and orchestration (Docker, Kubernetes).
• Experience with observability tooling (metrics, tracing, logging) for ML or distributed systems.
• Strong communication skills and comfort working cross functionally with data science, platform and product teams.
• Fluent English.
• Based in Europe.

• Exposure to LLM and agent frameworks and evaluation practices.
• Experience in a regulated or high throughput domain (fintech, payments, healthcare).
• Contributions to open source MLOps or agentic tooling.
• Experience with cloud ML platforms (SageMaker, Vertex AI, Databricks).

We are looking for a Machine Learning Engineer (mid level, 5 to 8 years of experience) based in Europe to build and scale the infrastructure that takes machine learning models from notebook to production. You will own the MLOps foundation, automate the model lifecycle, integrate streaming data pipelines and bring agentic capabilities into our ML systems. This is a hands on role for someone who cares as much about reliability and observability as about model performance. You will join the Data, AI and ML team and work closely with data science, platform and product teams.

• MLOps framework: design, build and maintain the MLOps platform (experiment tracking, model registry, versioning and reproducible training pipelines); establish CI/CD practices for ML with automated testing, validation gates and promotion workflows from dev to production; define standards and tooling for feature stores, model artifacts and environment reproducibility across teams.
• ML productionizing: take models from research or prototype stage to robust, scalable production services; build low latency, high availability serving infrastructure (batch, online and real time inference); implement monitoring for model performance, data drift and concept drift, with clear alerting and rollback paths; partner with data science teams to harden models for production constraints (latency, cost, scale).
• Automation of ML models: automate retraining, evaluation and deployment pipelines to reduce manual intervention; build self healing and auto rollback mechanisms triggered by performance or drift thresholds; create tooling that lets ML practitioners ship models without needing deep infra expertise.
• Streaming integration: integrate ML models with streaming data platforms (Kafka, Kinesis, Flink) for real time feature computation and inference; design low latency feature pipelines that bridge batch and streaming data sources; ensure consistency between offline (training) and online (serving) feature computation.
• Agentic integration for ML: design and integrate agentic workflows (LLM based agents, tool calling pipelines) alongside traditional ML models; build the observability, guardrails and evaluation frameworks needed to run agentic systems reliably in production; explore how agents can automate parts of the ML lifecycle itself (monitoring, triage, retraining decisions).