AI Systems Architect

About Marmalade
Marmalade is an invoice payments platform on a mission to revolutionise how businesses pay and get paid. We empower businesses to get their invoices paid when they need it, alleviating cash flow strain without debt and enabling them to grow on their terms.

The role
The AI Systems Architect is a senior role responsible for designing Marmalade's canonical AI architecture and the technical evaluation framework that keeps it safe. You will set the reference patterns, integration standards, and build-versus-buy calls that every AI capability is built against. You will translate the risk and governance frame into evals, guardrails, and incident response, and you will own the technical layer of Marmalade's AI controls library. The role sits in the Data, AI and Systems function.

What you'll do
You will architect the AI stack, set the standards, and own the technical evaluation and incident response capability that sits underneath every AI deployment at Marmalade.

Key Responsibilities
Systems Design

• Design Marmalade's canonical AI architecture against the data and integration plan: data schemas, contracts and custom agents.
• Design conventions, integration standards and build-versus-buy recommendations; surface the trade-offs for sign-off.
• Lead technical evaluation of model and data vendors and feed findings up the chain-of-command.
• Take ownership of our infrastructure including AWS and Data stores (PostgreSQL, Redshift, Salesforce)

Safety, Evaluation and Guardrails

• Translate the risk and governance frame into the technical evaluation framework: offline evals, online metrics, regression suites.
• Implement fintech-grade guardrails: PII handling, prompt-injection defence, audit trails.
• Own the technical layer of the AI controls library.

Incident Response and Continuity

• Define the AI incident response playbook within the risk framework: detection thresholds, severity classification, escalation paths.
• Own the post-incident learning loop and feed improvements back into architecture and standards.
• Direct AI developers in execution against the playbook.

Your background
You have:

• 10+ years of software and systems engineering experience, with significant time spent designing production systems end-to-end.
• Fluency in LLM application architecture: retrieval, agents, tool use, model serving, and the trade-offs between them.
• Strong AWS experience and a working command of Infrastructure as Code (Terraform).
• Hands-on capability across ETL and data pipelines, PostgreSQL and at least one production application framework (Ruby on Rails an advantage).
• Experience designing architectures and integration standards that other engineers build against.
• Track record of building offline and online evaluation frameworks for ML or LLM systems, not just shipping models.
• Experience implementing fintech-grade controls: PII handling, prompt-injection defence, audit trails, regression suites.
• Demonstrated incident response leadership: detection, severity classification, escalation, post-incident learning.

Proven Success in:

• Designing canonical architectures that survived contact with production.
• Leading technical vendor evaluation and feeding build-versus-buy recommendations into senior decision-making.
• Working in financial services or another regulated, risk-sensitive domain.
• Domain-Driven Design and risk and controls literacy.

What we offer

• Competitive salary and benefits. $200k base.
• The rare opportunity to build an AI-native fintech stack from the ground up at one of Australia's most innovative fintechs.
• Access to best-in-class AI tooling. We invest in the stack so you can move fast.
• Clear development and career progression opportunities.
• In-person role: Monday to Thursday in the office and Friday WFH.
• Swanky Melbourne CBD office with a vibrant and social work culture.