What is the most pythonic way to handle ElasticSearch?

I have an existing repository connected to MySQL. Details have been omitted for clarity.

class Person(object):   
def __init__(self):
    # This is entity is mapped to the DB using SQLAlchemy's classical mapping.
    It also includes an Address
    pass

class Address(object):
    def __init__(self):
        pass

class PersonRepository(object):
     def __init__(self, DBSession):
         self.session = DBSession

     def persist(self, entity):
         self.session.add(entity)
         self.session.flush()
         return entity  

 class AddressEditor(object):
        def __init__(self):
            pass

    def add_address(self, person, address):
        person.address = address
        #We can either inject the session into this method, or 
        #obtain it from a configuration file.
        repository = PersonRepository(DBSession)
        person = repository.persist(person)
        return person.address

I want to convert the repository to ElasticSearch. What is the best way to move forward?