This is a pretty ambitious project. What part are you starting with first — the SQL parser/planner or the storage engine?
Also curious what you mean by “little twists”. Are you aiming for PostgreSQL compatibility, or mainly using it to explore different design choices around things like MVCC and WAL?