Distributed DB Systems Engineer (Madrid Hybrid)

FastS3, Inc.; Madrid; Presencial; EUR 70.000 - 100.000; Jornada completa

Descripción de la vacante

FastS3 is building AI-native data infrastructure and seeks a Database Systems Engineer to design the database layer above PixelDB and work with TEMPANO for Iceberg-like cataloging.

This role targets a distributed, transactional, analytical, and vector workloads-capable system. You will prototype core components, refine commit logs, logging of transactions, and integrate with Iceberg-style metadata while maintaining performance.

Formación

• Strong systems programming experience in Rust, C++, Go, or C.
• Practical knowledge of database internals, storage engines, distributed databases, or query engines.
• Experience with MVCC, WAL/commit logs, transaction processing, indexing, compaction, or recovery.
• Understanding of distributed systems concepts such as consensus, replication, failure recovery, consistency, and concurrency control.

Responsabilidades

• Design and prototype a transactional table engine on top of PixelDB.
• Work on MVCC, commit logs, transaction visibility, snapshot isolation, and serializable consistency.
• Design stable Row ID abstractions so indexes point to logical rows rather than physical storage locations.
• Build metadata and manifest flows that connect live transactional writes to TEMPANO and Iceberg-compatible analytical snapshots.
• Help design compaction, version cleanup, and garbage collection that do not block long-running analytical queries.
• Explore query execution paths for different workloads including vector retrieval.

Conocimientos

Rust

C++

Go

C

Descripción del empleo

FastS3 is building AI-native data infrastructure and seeks a Database Systems Engineer to design the database layer above PixelDB and work with TEMPANO for Iceberg-like cataloging.

This role targets a distributed, transactional, analytical, and vector workloads-capable system. You will prototype core components, refine commit logs, logging of transactions, and integrate with Iceberg-style metadata while maintaining performance.