turbopuffer is looking for database engineers who want to advance the frontier of search.
what you'll do:
- design and implement improvements across one or more layers of our search engine (storage, indexing, querying)
- intensive performance hacking: tracing, CPU/heap profiling, optimizing data structures, using SIMD, tuning IO pipelines, etc.
- evolve our search primitives and APIs (full text + vector + reranking)
- work closely with customers to win their workloads and fold learnings back into the product
- write crisp docs
must have:
- systems engineering
- you've worked extensively in a systems language (Rust, C++, Java, Go, etc.)
- you think in memory layouts, cache lines, and syscalls
- you methodically work through problems until you've found the root cause
- domain expertise — you're an expert in at least one of the following domains
- search internals (e.g., inverted indexes, ANN indexes, rerankers)
- storage engines (e.g., LSMs, WALs, MVCC, compaction, GC)
- distributed systems (e.g., replication, leases, fault injection, simulation/chaos testing)
- linux — you're at home in terminals, eBPF/perf, strace, gdb
- typist — you can write well and explain complex ideas simply
- human — you build trust and admit what you don’t know
nice to have:
- multiple years of experience writing Rust
- you've maintained or operated serious systems in production, at scale
- you've made interesting contributions to the databases, IR, or systems communities (e.g., research, open source, speaking)
- additional domain expertise in:
- k8s and cloud infra (GKE/EKS/AKS, disks and networking, load balancers, observability)
- security (multi-tenant systems, isolation, sandboxing)
values we live by:
- overstep > understep
- correctness > simplicity > performance
- customer traces > first principles > hunches
- show > tell