Jun 2025 – Present · 11 mos

• SlateDB is a cloud-native embedded storage engine built on a log-structured merge-tree (LSM-tree) architecture. Unlike traditional LSM-tree engines, SlateDB stores data in object storage systems (e.g., S3, GCS, ABS, MinIO, Tigris), enabling bottomless storage, high durability, and simplified replication. This design trades off increased latency and API costs for cloud-native scalability.
• Current Focus:
• Implementing persistent compaction metadata to support reliable retries of failed or interrupted compactions.
• Developing a state management protocol to keep manifest and compaction files in sync, ensuring metadata consistency across operations.
• Addressing garbage collection edge cases to improve correctness and prevent resource leaks.
• 📄 Authored RFC 0013: Compaction State Persistence, outlining a durable approach to compaction state management.