DynamoDB Access Pattern

Question

I'm trying to familiarize myself with DynamoDB and I'm facing the following problem. My data record looks like this:

Crew{
  id: string,
  firstName: string,
  lastName: string,
  email: string,
  roles: Role[],
  address: Address
}

Role{
  name: string
}

Address{
  street: string,
  zipCode: string,
  city: string,
  countryIsoCode: string
}

In DynamoDB, it would look like this:

PK	SK	GSI_PK1	GSI_SK1	GSI_PK2	GSI_SK2	Entity	Data
crewID#1	crew				Crew		{firstName: 'Lex', lastName:'Luther', email: 'lext_luther@test.com', roles:[actor, director, writer]}
crewID#2	crew				Crew		{firstName: 'Bruce', lastName:'Wayne', email: 'batman@test.com', roles:[actor]}
crewReqID#1	req			CrewReq	CrewReq#2023-06-28-18:00:05		{title: 'Batman Begins', shootingStart:'fall',crewReq:[ crewID#2, crewID#10]}

I have the following access patterns that I need to solve without using a scan or multiple queries:

1. I want to retrieve all Crew items that have roles x, y, or n.
2. I want to retrieve all Crew items that have roles x, y, or n but do not appear in the crewReq attribute of crewReq#.
3. I want to retrieve a specific crew item.

I thought about removing the 'roles' attribute from the 'Data' field and writing the item to the DB for each role once. This way, I could at least solve the first and third access pattern in a single query. However, I'm not sure how to solve the second requirement yet.

Moreover, it seems like this approach might be incorrect. I also believe that a third GSI (Global Secondary Index) is unnecessary.

Do you have any ideas or suggestions?