Querying vs Scanning practice for specific scenario DynamoDB

There seems to be about a hundred ways of retrieving items in dynamo db. I have heard that Querying is only for Hash and Range Key values, however the, I have also seen that scanning is very slow and expensive on memory for my read and write units. However, some of the code and articles I've read are outdated, I am unsure where to go from here.

I need some help determining which is the most efficient way to find entries in the following ways.

Currently all my data is contained on one table "User_Table", I have "username" as my hash key, but do not use a Range Key, the items I'm trying to load should be around 30-50kb each

Scenario #1 Search for accounts with 2 attributes between 2 values each

I want to find nearby users locations so I can add them to a GoogleMap on the home screen, so I need to search dynamodb for my users with a "latitude" value between A and B, also a "longitude" value between C and D.

Scenario #2 Search by Hash Key

A user wants to search for others by username

Scenario #3 searching for a specific non hash key value (string)

My app allows users to add for example a Twitter account to their profile, no other profiles are allowed to have the same Twitter account linked, so I would just need to search and find the user containing a "twitter" value of "billymitchell" etc

QUERYING

querying has these different options

dynamoDBMapper.query(), dynamoDBMapper.queryPage() 
amazonDynamoDBClient.query(), amazonDynamoDBClient.getItem(), amazonDynamoDBClient.batchGetItem()

SCANNING

scanning seems to have the following options

dynamoDBMapper.scan(), dynamoDBMapper.scanPage(), dynamoDBMapper.parallelScan() 
amazonDynamoDBClient.scan()

I only have a few a entries in my DynamoDB Table currently, so I don't have a clear view of how well things will scale if my app starts to gain thousands of requests at a time, also . I know that adding more read and write units can get expensive very quick. Can anyone point me in the right direction?

Thanks!