ML Platform Engineer — Scale Foundation-Model Infra

kaiko.ai; Zürich; Hybrid; CHF 120.000 - 180.000; Vollzeit

Benefits dieser Stelle

Competitive salary

Pension plan

25 vacation days

Offsites & events

Learning budget €1000

Commuting subsidy

Zusammenfassung

kaiko is building a next-generation clinical AI assistant to help clinicians reason across patient data, guidelines, and diagnostics. Our platform scales ML workflows from training to deployment, with a focus on reliability, safety, and regulatory alignment.

As an ML Platform Engineer, you will shape the foundation used for foundation-model training and serving, collaborating with research and product teams to deliver scalable, observable infrastructure.

Qualifikationen

• 2–5 years in production ML platform engineering or ML Ops.
• Hands-on with Kubernetes, Helm, Terraform, Docker and CI/CD tools.
• Experience scheduling GPU workloads on Kubernetes, Ray, or similar.
• Familiar with the full ML workflow: training, experiments, registry and model serving.
• Strong software engineering with Python; Go or other languages welcome.

Aufgaben

• Build and evolve infrastructure for fast, reliable ML development.
• Scale GPU workloads across on‑prem and cloud clusters.
• Support hybrid and multi‑cloud strategies including storage rollout.
• Own parts of the AI Factory and orchestration (Dagster, Ray).
• Develop model lifecycle tooling: tracking, registry, versioning, serving.
• Collaborate with research and product to translate needs into shared infra.
• Maintain lineage, reproducibility, and clear documentation.

Kenntnisse

Kubernetes

Helm

Terraform

Docker

CI/CD tooling

Observability

Python

Tools

Dagster

Ray

KAI-Scheduler

Hammerspace

Jobbeschreibung

kaiko is building a next-generation clinical AI assistant to help clinicians reason across patient data, guidelines, and diagnostics. Our platform scales ML workflows from training to deployment, with a focus on reliability, safety, and regulatory alignment.

As an ML Platform Engineer, you will shape the foundation used for foundation-model training and serving, collaborating with research and product teams to deliver scalable, observable infrastructure.