Senior Software Engineer (YDB Team)

About Nebius: Nebius is leading a new era in cloud infrastructure for the global AI economy. We are building a full-stack AI cloud platform that supports developers and enterprises from data and model training through to production deployment, without the cost and complexity of building large in-house AI/ML infrastructure. Built by engineers, for engineers. From large-scale GPU orchestration to inference optimization, we own the hard problems across compute, storage, networking and applied AI. Listed on Nasdaq (NBIS) and headquartered in Amsterdam, we have a global footprint with R&D hubs across Europe, the UK, North America and Israel. Our team of 1,500+ includes hundreds of engineers with deep expertise across hardware, software and AI R&D. The role The YDB Development Team at Nebius is hiring mid- and senior-level software developers with C++ expertise. YDB is an open-source platform with two roles within Nebius: Providing the binary object storage foundation for S3 Object Storage, Shared Filesystems and Network Block Storage Serving as a distributed relational database and message broker for services such as IAM, billing, audit trails and logging We operate YDB clusters in more than 10 regions using HDD and NVMe storage. Individual clusters span tens of thousands of nodes, manage hundreds of petabytes of data and deliver terabytes per second of throughput. Our infrastructure includes x86 and Arm-based systems, including NVIDIA GB300 nodes. What you’ll work on Evolve YDB’s architecture for new hardware, redesigning core components when incremental optimization is insufficient: QLC NVMe drives: increase storage density while controlling RAM/CPU consumption and write amplification Dual 800 Gb/s network adapters: reduce CPU overhead and memory bandwidth consumption to utilize the available NIC bandwidth NVIDIA GB300-based systems: ensure efficient operation on Arm-based hosts and offload storage and networking processing to BlueField DPUs Support high-capacity SMR HDDs by designing data placement, sequential writes, garbage collection and recovery around zoned-write constraints Develop wide erasure-coding schemes to reduce space aplimfication while maintaining durability and availability Reduce recovery and rebalancing times after device outages while limiting the impact on foreground workloads Prevent, detect and recover from silent data corruption across memory, networking and storage Provide the system capabilities and interfaces required by fleet-wide automation so that hundreds of production clusters can be deployed, configured, upgraded, diagnosed and operated consistently We move quickly and focus on delivering practical value. Features and optimizations merged into the production branch are typically deployed within weeks. You’ll see your work solving real problems in live clusters—not waiting years to make an impact. Requirements: 5+ years of professional experience developing high-load systems in C or C++. Candidates with 3+ years will also be considered if their experience includes relevant work on distributed storage systems Experience developing high-performance, low-latency software components Understanding of core dumps, flame graphs and sanitizer-enabled builds Understanding of CPU caches, atomic operations and NUMA architectures Nice to have: Experience designing and developing production-grade distributed storage systems and/or databases Experience with on-disk data structures such as LSM trees and B+ trees, and storage algorithms such as erasure coding and checksumming Understanding of NVMe, conventional HDD and SMR HDD internals Experience with debugging and profiling tools such as perf, VTune, bpftrace and gdb Familiarity with the Linux kernel, SPDK/DPDK, libaio and io_uring Knowledge of IP, TCP, UDP and DNS, and ideally InfiniBand or RoCE/RDMA Experience with Kubernetes and Grafana Experience responding to production incidents Knowledge of availability and durability calculations