Senior AI Platform Engineer

TMS Technology; Zürich; On-site; CHF 120,000 - 190,000; Full time

Job summary

TMS Technology is hiring for a role that turns data science models into production systems capable of running inside client environments, including air-gapped banks. You will own the end-to-end deployment platforms, from training pipelines to serving layers, ensuring reliability in highly regulated infrastructure.

Deep Kubernetes and distributed systems expertise, plus hands-on MLflow and Spark/vLLM, are needed.

Qualifications

• Proven experience deploying AI/ML platforms into restricted or isolated environments.
• Strong Kubernetes and distributed systems capabilities in production.
• Hands-on with MLOps tooling (MLflow, equivalents) and Spark or vLLM.
• Ability to explain technical decisions to client infrastructure teams.

Responsibilities

• Build frameworks and pipelines to run AI/ML models in client environments with limited or no outbound internet.
• Deploy and operate ML platforms in isolated, air-gapped settings and ensure production durability.
• Integrate ML systems with Kafka and gRPC-based backends and collaborate with infra and software teams.
• Maintain secure, scalable, and maintainable production deployments and respond to incidents.

Skills

Kubernetes

MLOps

Spark

Kafka

gRPC

MLflow

Distributed systems

Helm

Tools

vLLM

Job description

A Swiss RegTech company whose platform fights financial crime inside more than a thousand banks across eighty countries, an AIFinTech100 member, is building out the function that turns its data science team’s models into software that actually survives contact with a bank’s infrastructure.

That’s the job. Their data scientists build the models, fraud detection, AML scoring, sanctions screening. What happens next, the Kubernetes platform, the MLOps pipelines, the serving layer that has to run inside client environments that are often isolated, air-gapped, or running on whatever GPU happens to be sitting in that bank’s rack, is yours to own.

This isn’t a cloud-native MLOps role where everything lives in a managed service and scales on demand. Some of your deployments will have no outbound internet at all. You’ll be building the frameworks and pipelines that make model training, testing, deployment and monitoring repeatable across very different customer environments, integrating them into Kafka and gRPC-based backend systems, and working directly with the infrastructure and software engineering teams to get it all into production, not around them.

What matters most:

• You’ve deployed AI/ML platforms into isolated, restricted, or non-SaaS environments, ideally in live production, not just in a lab or a PoC
• Deep Kubernetes and distributed systems experience, with MLOps tooling (MLflow and equivalents) and a framework like Spark or vLLM in your actual production toolkit
• You write clean, secure, maintainable code because you’re the one who gets paged when it breaks
• You’re as comfortable explaining a technical decision to a client’s infrastructure team as you are writing the Helm chart yourself

Financial crime domain knowledge is a bonus, not a requirement. What we need is someone who’s done the unglamorous, high-stakes work of making AI run where the cloud can’t reach, and wants to keep doing exactly that, at scale, for a platform operating inside some of the most regulated infrastructure in the world.