Bitpanda uses a multi-cloud strategy with AWS as the primary provider to build for scale and avoid vendor lock-in.
A real-world example of migrating data from a failing AWS region in the UAE to Europe highlights the importance of disaster recovery planning.
The platform is designed for resilience using tools like Kafka to handle sudden 20x traffic spikes caused by external events.
Development teams own their domains completely, including on-call duties, which is managed by a central service management team for large incidents.
B2B and B2C offerings are treated as channels that consume a shared, configurable core platform for services like trading and hedging.
The tech stack includes AWS, Kubernetes, Java, Python, and TypeScript, pragmatically choosing the best language for each domain's needs.
AI tools like CloudCode are used to boost productivity, but the guiding principle is that humans must be able to understand and debug the code at 3 AM.
Endpoint security and tooling are used to prevent data leakage by treating AI agents as regular user actors and monitoring their activity.
Instead of unstructured '10% time', innovation is encouraged through hackathons focused on solving specific, existing business challenges.
The level of creative freedom varies by product area, with strict controls on critical systems like hedging to ensure compliance and protect customer funds.
A culture of proactive design via RFCs and reactive learning through blameless postmortems supports a fast and responsible continuous deployment pipeline.
A multi-layered security approach combines bug bounties, vulnerability scanning, and a strong focus on endpoint security and employee awareness to combat modern threats.
The focus is on creating a robust, scalable, and well-documented API layer that can be consumed equally by human-facing UIs or autonomous AI agents.
Jobs that call for the skills explored in this talk.