Random experiments that re-invent the wheel for the purpose of learning about wheels.
While the general idea was to implement a key-value store with log structured merge trees, in actuality it will probably just end up as a disconnected set of experiments for learning purposes.
P.S. If you landed here looking for a real implementation, try LevelDB instead.