As part of the Search, Recommendation, Ads (Engineering & Architecture) team, We build and operate the search engine that powers retrieval and ranking across the company. The system is built on Vespa (a ~1.7M-line C++/Java codebase) and runs at scale across hundreds of nodes. Our work spans the full stack: cluster orchestration, HNSW-based vector indexing, rank expression evaluation, capacity planning, and the developer tooling that keeps all of it shippable. We care a lot about correctness, latency tails, and operational simplicity.

• Design, implement, and operate components of the search engine — indexing, query execution, ranking, or storage using C++
• Improve indexing and retrieval performance: SIMD-friendly data layouts, cache-aware algorithms, lock contention
• Extend vector search capabilities (HNSW and beyond) and integrate them cleanly with traditional inverted-index retrieval
• Build tooli...