I think you'll have to give us more information. How long should this project take? At what level? For example I could suggest you implement your own RDMS but that might be way too long/difficult.
An auto-indexing system based on tracking of poorly performing queries, their explain plans, and what index will provide the most improvement for the smallest disk cost without over-specializing to a particular query. The optimizer itself can't negatively affect the performance of the system and the indexes must be added live.
An auto-indexing system based on tracking of poorly performing queries, their explain plans, and what index will provide the most improvement for the smallest disk cost without over-specializing to a particular query.
Well the "website connected to a database" is the least interesting part of what you have so far... the "smart" part is where you need to do some brainstorming.