Thanks for sending this through. We have a few suggestions around functionality that we've found very useful in the past, but otherwise it looks promising.

1. Query on keys or portion of keys

2. Ability to query on unindexed attributes

3. the ability to specify a custom filter is potentially useful to implement missing or business logic, although in practice you may be able to construct these with the given filters.
Example: Filter out currency values older than X if the time is between 10am and 11am

4.  The ability to specify a custom extractor which could be used to extract a portion of stored data or manipulate it before comparison.
Example : data is being stored as an array and we're interested in the Nth value in the array -> this is a requirement from a previous project

5. equality and range filters should also include the "andEquals" options: lessThanEquals, greaterThanEquals

6. all(List) and any(List) functions are easier to use than chaining together and() and or() statements

7. I'd suggest with() rather than having() as this is closer to SQL syntax, which has a different meaning for HAVING

Do queries require that the entire object be deserialised before the filter can be evaluated or do you compare on deserialised index values?
This is potentially quite slow, and requires that matching Java classes are provided on the server side even if the client is .NET

Rob