Apache Flink® PMC Member & Committer, specialized in state management and checkpointing — Flink's state backends, checkpoint coordination, savepoints, and recovery semantics.

Beyond Flink, I'm building my own open-source (Apache-2.0) storage and stream-processing stack in Rust: an LSM-based key-value storage engine designed for the distributed/cloud era, plus a stateful stream processing engine.

A high-performance LSM-based embedded key-value store for distributed systems and standalone apps — think RocksDB with distributed superpowers:

• Hybrid local/remote object storage (S3, OSS, …)
• Distributed global snapshots & remote compaction
• Key-value separation, TTL, hot/cold tiering
• Schema evolution & multi-version snapshots
• Multi-language bindings (Java supported)

A unified storage layer for Apache Flink that makes state and sink data disaggregated, observable, and consumable:

• Flink state backend + SQL source/sink
• Disaggregated state with high performance and fast rescaling
• Web monitor to inspect checkpoint & sink snapshots without touching running jobs

A stateful stream processing engine focused on rapid scaling, rolling updates, and record-level observability — stay tuned. 🔮