[Your Name] · [Email] · [Phone] · [City, ST]
April 21, 2026
Dear Hiring Manager,
I'm applying for the Senior Software Engineer, Search Infrastructure role. Your team's recent paper on query fan-out optimization in the serving path is directly adjacent to what I've spent the last three years on at LinkedIn — cutting tail latency for a federated search service that fans out across 14 backends — and the problem framing in the paper is what convinced me to apply now instead of waiting another cycle.
At LinkedIn I led the rewrite of our federated query router, which handles 380K QPS on the hot path and serves 1.1B monthly users. The system was burning 28% of its CPU on protobuf serialization across microservice hops; I proposed a partial move to a shared Arrow-based columnar format for fan-out results, prototyped it on a single query class to prove the numbers, and then drove the 9-team migration over two quarters. P99 latency dropped from 480ms to 170ms and we reclaimed 2,100 cores of capacity — roughly $4.2M in annual infra savings. The part I'm most proud of isn't the number: it's that the rollout plan made it safe for the 9 owning teams to say no, and two of them did, and the design held up anyway.
Before LinkedIn I spent four years at a search startup (Algolia, early London team), where I shipped the first multi-region replication feature and on-called for it for a year. I've published two papers at SIGMOD on approximate top-k retrieval, keep the retrieval-benchmarks open-source repo, and review NeurIPS submissions on efficient retrieval. I'd bring that combination — production-grade infrastructure plus a research instinct for why a system works — to a team that already operates at Google's scale.
I'd love a first conversation to understand where the biggest open problems on the serving path are, and to share a short design doc I wrote on the Arrow migration (internal-scrubbed) if the hiring committee finds it useful. I'm happy to do the coding rounds in C++, Go, or Python.
Sincerely,
[Your Name]