Software Engineer, Global Advertising Data Platform

Team Introduction The Global Advertising Data Platform team builds the data foundation and products behind advertising across our international products — petabyte-scale ingestion and storage through to the query and data products thousands of internal users and systems depend on every day. The team is dedicated to making an AI-native data platform that enables agents to be reliable colleagues across all data-related scenarios. Semantic modeling, compile-time governance, agent-facing services and tools, evaluation infrastructure, all at global advertising scale. The work is measured by what it delivers: a better advertiser experience, efficiency lift, measurable business growth, or strategy that lands quickly with the platform's help.

Responsibilities - Design and build the core of an AI-native data platform: Own architecture and hands-on development of high-throughput services serving hundreds of millions of users' advertising activity. You will take up versatile roles and deliver solutions end to end: Identify and define the problem, design technical solutions, build the system with agents, tune and optimize the outcome.

- Design agent-facing data services and tool interfaces: Build the APIs and tool surfaces (MCP-style typed interfaces, retrieval over catalog and metadata, structured query execution) through which agents discover and use data — with scoped permissions, typed inputs, and full audit trails, so capability and safety ship together.

- Contribute to the semantic layer that makes data agent-consumable: Turn tribal knowledge into machine-readable definitions: metrics, dimensions, lineage, valid join paths, and data contracts, etc. This is the core that decides whether an agent's data work is stable, reliable and verifiable.

- Make correctness measurable: Contribute to code and testing standards and to the quality control methods the team runs on — extended to a world where both humans and models write code and generate queries. Build golden query suites, regression and evaluation harnesses, lineage-based impact analysis, and data quality monitoring that catch semantic drift before users do.

- Conquer technical challenges: High concurrency, multi-tenant data isolation and governance enforced at query-generation time rather than patched afterwards, system decoupling, cost and performance at scale — these are the recurring challenges of the platform, and you'll work with the team to break through them.