About the Team:
The ByteDance Database LavaStore team builds and maintains ByteDance's purpose-built local storage engine for large-scale cloud services. Working at the intersection of storage engines, databases, filesystems, and performance engineering, the team solves highly write-intensive, latency-sensitive, and cost-sensitive infrastructure problems that general-purpose systems cannot fully address. The team combines systems research with end-to-end production ownership: it studies real workload patterns from database, key-value, and stream-processing systems, turns those insights into practical storage innovations, and validates them in production at massive scale. Its work spans LSM-based engine design, KV separation, garbage collection, WAL-oriented storage paths, append-optimized filesystems, caching, and tail-latency optimization. LavaStore has already been deployed at very large scale, with over 100,000 running instances, more than 100 PB of stored data, and over 2 billion requests per second, reflecting the team's focus on building simple, robust, and cost-effective infrastructure with measurable real-world impact.
Responsibilities:
- Participate in the design and implementation of high-performance KV storage engines and file systems for large-scale distributed storage & database system backends;
- Explore the design and evolution of next-generation KV storage engine architectures to ensure low latency and high throughput;
- Deeply understand storage requirements across business scenarios, and collaborate with business teams to find the most suitable storage solutions.Minimum Qualifications:
- Bachelor's degree or higher in Computer Science, Computer Engineering, Electrical Engineering or related majors;
- Strong problem analysis, conceptual abstraction and design skills; proficient in C/C++, familiar with at least one of Python / Shell or similar languages;
- Proficient in systems programming, including system calls and library functions, network/multithreaded programming models, debugging, and IO performance tuning tools and techniques;
- In-depth understanding of storage system architectures and principles, such as LSM Tree, file systems (ext4/bluefs), etc.;
- Solid understanding of distributed systems fundamentals, including consensus, replication, consistency models, and fault tolerance;
- Excellent ability to identify and solve problems, good communication skills, team collaboration spirit, and a strong desire to learn new technologies.
Preferred Qualifications:
- PhD 's degree or higher in Computer Science, Computer Engineering, Electrical Engineering or related majors;
- 5 years experience with asynchronous programming models and event-driven architectures, including callbacks, futures/promises, and coroutines;
- Prior development experience in any of the following: large-scale distributed storage / caching / database systems / single-node KV storage engines / single-node file systems;
- Familiarity with one or more common open-source projects such as RocksDB / LevelDB / Redis / Memcached / Cachelib; Prior RocksDB development experience is a plus;
- Prior development experience with storage-compute disaggregation architectures is a plus.