Check out my latest posts!


I've been deep in the trenches building SmolDB, a C++20 mini-RDBMS from scratch. The project's goal is to create a database optimized for a specific workload: high-concurrency transactions on relatively small tables, similar to what you might find in a core banking system. This philosophy—high throu...
Designing a High Performance RDBMS Buffer Pool

Designing a High Performance RDBMS Buffer Pool

Blog
Distributed databases promise linear scaling in both data size and throughput (at least in theory). In practice, harnessing the full power of a distributed system, especially one based on PostgreSQL like Citus, demands a holistic approach. While advanced partitioning schemes, elaborate caching layer...
Tuning Distributed PostgreSQL

Tuning Distributed PostgreSQL

Blog
# First Pass Systematic testing of concurrent programs has been researched for quite some time and it is known to have the problem of *state-space explosion* whereby testing all possible interleaving of concurrent programs is exponential in the execution length. Deterministic Partial Order Reduction...
Partial Order Aware Concurrency Sampling (POS) - Yuan, et al.

Partial Order Aware Concurrency Sampling (POS) - Yuan, et al.

PaperBlog
# Problem Even when requirements are clear, verifying software correctness and ensuring our code works in all scenarios without affecting existing logic can be challenging. This is more apparent on ‘hot’ codes: Excerpts of code that got updated often due to its status as a new base logic or due to ...
Improving Software Testability with Math

Improving Software Testability with Math

Blog