Senior Software Engineer, Data Platform & AI Enablement

What matters most

• Architecting High-Performance Distributed Systems

  The role requires leading the design and implementation of core infrastructure capable of processing petabytes of data across 30+ global regions. You must demonstrate the ability to move beyond standard pipelines to build custom, scalable storage and service-layer interfaces that support real-time financial data and AI workloads.

• Systems Internals and Performance Tuning

  A primary focus is deep-diving into the internals of distributed systems, specifically state management, checkpointing, and exactly-once semantics. You will be expected to optimize JVM performance and ensure millisecond-level accuracy for data processing at a global scale.

• Developing the Knowledge Platform for AI

  You will own the development of a platform that enables AI-driven products, specifically focusing on scalable vector indexing, agentic workflows, and real-time data streaming integration. This requires bridging the gap between raw data infrastructure and actionable AI capabilities.

• Technical Leadership and Engineering Rigor

  Beyond individual contribution, you are expected to advocate for elite engineering practices, contribute to shared SDKs and automation, and mentor mid-level engineers. This involves driving project delivery through design reviews and maintaining high standards across the engineering organization.

Likely challenges to think through

• Balancing Global Scale with Millisecond Accuracy

  Maintaining exactly-once semantics and low-latency performance while processing petabytes of data across 30+ regions is a significant technical hurdle. Prepare by reviewing strategies for distributed state management and the trade-offs between consistency and availability in high-throughput streaming systems.

• 0-1 Ownership in a Complex Infrastructure Environment

  The role demands taking full ownership of complex, ambiguous technical challenges from design to production. You should be prepared to discuss specific instances where you identified a foundational infrastructure gap, designed a custom solution, and navigated the operational complexities of deploying it at scale.

• Integrating AI Workflows into Legacy Financial Data

  Building a 'Knowledge Platform' that supports agentic workflows and vector indexing requires integrating modern AI paradigms with established, high-stakes financial data infrastructure. Prepare by considering how to architect modular systems that allow for rapid AI experimentation without compromising the stability or regulatory compliance of the core financial backbone.

Topics to prepare

• Distributed Systems Semantics

  Review the conceptual requirements for exactly-once processing, state management, and checkpointing in stream processing frameworks. Exercise: Design a theoretical architecture for a distributed system that guarantees exactly-once delivery across multiple global regions, documenting how you would handle network partitions and node failures.

• Lakehouse and Vector Indexing Architectures

  Study the integration of data lakehouse patterns with vector databases to support AI-driven workflows. Exercise: Explain the trade-offs between different vector indexing strategies (e.g., HNSW vs. IVF) in the context of real-time retrieval-augmented generation (RAG) systems.

Skills to revise

• JVM Performance Tuning

  Focus on memory management and garbage collection optimization for high-throughput Java/Kotlin applications. Exercise: Analyze a hypothetical scenario involving high latency in a Kafka-based stream processor and identify which JVM metrics (e.g., GC pause times, heap usage) would be critical to diagnose the bottleneck.

• Stream Processing Internals

  Deepen hands-on knowledge of Flink or Spark internals, specifically regarding stateful stream processing. Exercise: Write a code snippet or architectural diagram demonstrating how to manage large-scale state in Flink while maintaining low-latency performance during checkpointing.

• Distributed Infrastructure Design

  Practice designing scalable infrastructure components using Kubernetes and distributed messaging systems. Exercise: Propose a design for a service-layer interface that handles petabyte-scale data ingestion, ensuring high availability and fault tolerance across 30+ regions.

• Prepare a 0-1 project case study

  Select a complex distributed system project you led from inception to production. Be prepared to explain your architectural choices, how you handled state management or exactly-once semantics, and the specific performance trade-offs you made.

• Review distributed systems internals

  Refresh your knowledge on low-level systems concepts relevant to Kafka, Flink, or Spark. Focus on how these tools manage memory, networking protocols, and checkpointing to ensure high availability and data consistency at scale.

• Draft a technical design for AI infrastructure

  Create a high-level design document for a scalable vector indexing or real-time data streaming pipeline. This demonstrates your ability to apply first-principles thinking to modern AI enablement challenges.

• Prepare examples of JVM performance tuning

  Identify specific instances where you optimized Java or Kotlin applications. Be ready to discuss how you diagnosed memory leaks, garbage collection issues, or latency bottlenecks in a high-throughput environment.

• Formulate questions on engineering culture

  Prepare thoughtful questions about Airwallex's engineering practices, such as how they balance rapid feature delivery with the need for robust, petabyte-scale infrastructure, or how they approach mentorship for mid-level engineers.