ML Platform Engineer — Payments

Amelia Hughes

Full-time · Mid-level · Berlin

This role is no longer accepting applications

Similar open roles you can apply to now:

About the role

We are a payments infrastructure company. We process card-not-present transactions for 3,800 merchants across Europe, with a combined volume of €2.1 billion per month. Uptime is not a value here — it is a contractual obligation. Our SLA is 99.99%. The ML systems in our stack are not experimental: they power fraud detection, chargeback prediction, and dynamic 3DS authentication decisioning. These systems are in the critical path of every transaction we process. We are looking for an ML Platform Engineer to join the team that owns this infrastructure. The work involves Kubernetes-based model serving, real-time feature pipelines over Kafka, and the deployment tooling that allows our fraud science team to ship model updates without introducing regressions. The culture is engineering-first, detail-oriented, and deeply intolerant of the kind of technical debt that creates 3am incidents. If that sounds like a place where you would do your best work, we would like to talk to you.

Responsibilities

Own and maintain the real-time model serving infrastructure for fraud detection and authentication decisioning models
Build and improve the Kafka-based feature pipeline serving transaction risk models with sub-50ms SLA
Design and implement deployment tooling that allows the fraud science team to ship model updates with canary rollout and automatic rollback
Define and track ML infrastructure SLOs — latency percentiles, feature freshness, and prediction error rates — with automated alerting
Work with the fraud science team to establish deployment standards and reduce mean time from model approval to production

Requirements

4+ years of platform, MLOps, or data engineering with production experience in a real-time, latency-sensitive environment
Kubernetes for ML model serving — you understand horizontal pod autoscaling, resource limits, and liveness probes for serving workloads with sub-100ms SLAs
Kafka for real-time feature pipeline design — consumer group management, partition strategy, and exactly-once delivery guarantees
Python for infrastructure tooling, deployment automation, and data quality validation
MLflow or equivalent for model registry, versioning, and serving lifecycle management
Docker for reproducible model serving containers and consistent CI/CD deployment
SQL for feature store queries and model evaluation data extraction

Benefits

$95,000 – $117,000 base salary
Full remote — our engineering team is distributed across Germany, Netherlands, Spain, and Poland
25 days annual leave plus public holidays in your jurisdiction
$1,500 annual professional development budget
On-call is structured and compensated — one week in eight with clear escalation paths and post-incident reviews