About the role
You will work on AccountsDB, the custom storage engine at the heart of every Solana validator. You'll design, implement and optimize storage paths, indexing and caching layers, reduce memory and disk footprint, improve background services and snapshot handling, ensure state correctness across forks and crash recovery, and build thorough tests while collaborating with runtime, block production and protocol engineers.
Responsibilities
- Design, implement and optimize AccountsDB core storage paths including account file layout, account index and read/write caching layers
- Lower memory and disk requirements as account state grows
- Improve flushing, cleaning, shrinking and purging processes to reduce write amplification
- Optimize snapshot generation, loading and index generation to shorten startup and catch-up times
- Ensure account state remains correct across forks, rooting and crash recovery
- Create and execute stress tests, fault injection and performance benchmarks
- Collaborate with runtime, block production and core protocol engineers and other Solana client teams
- Maintain AccountsDB documentation and conduct peer code reviews
Requirements
- Bachelor's degree in CS, Engineering, or relevant practical experience
- 3+ years architecting data stores, file systems, databases, or high-throughput storage systems
- Mastery of Rust, C, or C++
- Knowledge of low-level storage mechanics including memory-mapped files, page caching, index structures, on-disk formats and garbage collection
- Advanced knowledge of kernel-level and hardware I/O, NVMe, buffered vs direct I/O and io_uring
- Understanding of custom memory allocators and Linux kernel scheduling and page management
- Experience building concurrent systems, lock-free structures and hardware memory models
- Experience profiling and tuning low-level code
- Strong debugging and problem-solving skills
- Excellent communication skills
Requirements
- Bachelor's degree in CS, Engineering, or relevant practical experience
- 3+ years architecting data stores, file systems, databases, or high-throughput storage systems
- Mastery of Rust, C, or C++
- Knowledge of low-level storage mechanics including memory-mapped files, page caching, index structures, on-disk formats and garbage collection