Senior Site Reliability Engineer – ML Platform

About the role

As a Senior Site Reliability Engineer you will own the reliability, scalability and observability of Vinted’s ML Platform services, working across on‑prem and cloud environments. You will partner with data infrastructure, production engineering and ML teams to deliver self‑service, low‑latency platform capabilities for feature stores, embedding pipelines and dataset tooling.

Key responsibilities

• Maintain and improve the Vinted Feature Store to meet strict latency SLOs of 400k+ reads per second.
• Evaluate, benchmark and prototype next‑generation distributed storage technologies as potential Vespa replacements.
• Build and automate self‑service embedding extraction pipelines, abstracting compute and serving complexity.
• Scale and operate the Dataset & Drift Platform for dataset discovery, versioning and drift monitoring.
• Lead reliability work: define SLOs, create alerting, write runbooks, conduct post‑mortems and reduce toil.
• Implement automated data governance to ensure GDPR‑compliant feature and dataset lifecycles.

Required profile

• Strong understanding of Linux operating systems and distributed systems architecture.
• Proven experience with core SRE practices (SLOs, observability, incident management) and a platform mindset.
• Excellent written and spoken English with strong cross‑team consulting and communication skills.

Required skills

• Linux
• Kubernetes
• Terraform
• Chef
• Google Cloud Platform
• Go (or other programming/scripting languages)
• Kafka
• Vespa
• Flink
• Temporal
• Vitess

What we offer

• Share options programme
• 25 days of holiday
• Home‑office equipment and up to €540 budget for furniture
• Private health insurance and mental‑health support via Mindletic
• Team‑building events and monthly Vinted shopping budget
• Workation option: up to 90 days per year, 21 outside the EU
• Dog‑friendly offices, gym & in‑house meals in Vilnius, lunch allowance in Kaunas
• Individual learning budget for courses, certifications and workshops
• Hybrid work model (recommended 2 days in office)