Senior Platform Engineer (Agentic AI Platform)
Bankdata
Required skills
About the role
Translated into English from Danish by Korum. The employer's original wording is on their site.
The banking sector is in the middle of a significant shift. Rising customer demands and complex operations challenge the traditional way of delivering advice and service. At the same time, much advisory work is based on patterns and data - and this is where artificial intelligence can help reduce costs, improve customer relationships and make banks more competitive.
Over the last few years, Bankdata has built experience with both classic Machine Learning, LLM-based search in business processes and AI agents for more complex tasks and hyper-personalised solutions. This has laid a solid foundation for our further work.
Now we are ready for the next step. The ambition is clear: artificial intelligence is a central part of Bankdata's offering to our customers. That is why we are building Bankdata's Agent Platform - the shared platform for developing, deploying, operating and governing AI agents across Bankdata and the banks.
Just as our container platform provides a standardised execution environment for applications, the Agent Platform must provide a standardised execution environment for AI agents. And we are not starting from scratch: in the Horizon platform, Bankdata already has a strong concept for container execution, which large parts of the business run on. That foundation - and the platform competences built up around it - is what we are taking forward. The task now is to lift it to workloads that behave differently from classic applications and that require their own guardrails.
Centralise the complexity, decentralise the value creation. Security, quality, compliance, IAM, explainability, traceability and input/output control belong to the platform. Intelligence, integration and business logic belong to the person building the agent.
Role: From golden path to self-service productionYou will be a key profile in making it easy, secure and scalable to build and run agents.
This means that you will not only work with LLMs and agent frameworks, but to a large extent with what makes solutions sustainable: infrastructure, automation, identity, observability and developer experience.
Agent workloads are not stateless services: they are non-deterministic, they cost money per call, they need their own identity, they hold state across restarts, and they stop and wait for a human. The platform must be built for that.
You will:
Build golden paths rather than tailor-made solutions per team, and automate provisioning, onboarding, compliance checks and telemetry
Build security and compliance into the platform, so that every agent inherits them rather than each team implementing its own
Ensure hard isolation between institutes, so that one tenant can never affect another's agents or data - and design the blast radius so that a fault in one place stays in one place
Develop the platform's agent primitives: a hosted tool-server layer with authentication, authorisation, rate limiting and audit, policy as code, tracing and structured logging, sandboxed execution, and controls for prompt injection, PII and content filtering
Build the approval mechanism, where an agent waits when a human must decide - so that human-in-the-loop is a platform capability rather than something every agent invents
Establish cost attribution per team, agent and use case - including token and inference costs
Work with agent identity and delegated authorisation, as well as session and state persistence, so that an agent's work survives a restart
You will work across the stack but with a clear emphasis on platform, infrastructure and developer experience.
You have helped build and launch the first golden path, which other teams actually build their agents on
At least one agent is in production with real users, without the team behind it having to build security, IAM, audit and traceability itself
Deployment of agents has become self-service - most deployments do not require help from the platform team
Observability, resilience and compliance come by default rather than as an afterthought
Cost attribution per team, agent and use case is in place, so that consumption can be explained and managed
You will become part of a new platform team building Bankdata's Agent Platform. The team is being set up and established on an ongoing basis as we implement the platform. We are building it in close collaboration with our banks - and the platform is not only for Bankdata's own developers. A bank must be able to build and put agents into production on equal terms with a Bankdata developer, without needing us in the loop.
This is a platform team - not an application team. We build capabilities that enable many other teams to build AI-driven products.
Because the technology landscape is changing rapidly, we are not looking for anyone who has 'done this before'. We are looking for experienced engineers who want to help define how it should be done.
There is great responsibility for your own work and a focus on progress. Sparring takes place through pair work, workshops and ongoing technical discussions.
The profile we are looking for: Platform engineer who thinks of platform as productWe are looking for you who know that an agent in production is far more than a model - the platform around the model is what must be built once and work every time.
You have experience with:
Cloud platforms, microservices and distributed systems
Kubernetes, container platforms, multi-tenancy and modern platform engineering
CI/CD, infrastructure as code and automation
Observability, scalability, resilience, networking and identity
Software development and API design in Java, Python or .NET
GenAI components such as LLMs, RAG or agent systems - or a strong drive to learn them quickly
You think in reusable platform capabilities rather than single solutions, and you reduce the cognitive load for the teams building on top. You shift left on requirements from legislation and compliance, so that they are built into the paved path rather than checked afterwards.
We are open to both experienced profiles and promising developers who are early in their careers. What matters most is that you can build - and take responsibility.
You thrive when you:
Take ownership rather than wait for direction
Find solutions in complexity and uncertainty
Prioritise progress without compromising on quality
would rather solve the problem than analyse why it is difficult
You will get the opportunity to:
Take greater technical responsibility in the team
Drive solutions in close collaboration with the banks and the business domains
Help define architecture and standards for agents in production across the community
If you want to work with GenAI in a context where it must work in reality - not just in a demo - then let us have a conversation.
Bankdata builds software for the financial sector. We are almost 1,100 colleagues on the team, and we develop digital products, platforms and infrastructure that roughly every third Dane, 200,000 businesses and several thousand bank employees use every day.
We believe in structured personal development, a strong collegial community, and that ambitious employees with flexible and sustainable working lives perform better. Learn more about working life at Bankdata here.
At Bankdata we work purposefully to promote diversity among our employees, and we therefore encourage everyone, regardless of gender identity, age and cultural background, to apply for jobs with us.
See how well your CV fits this role. Log in or create a free account to check.