Dumb, straightforward seeding Sequelize data. Feel free to improve. :-)

seed.js

#! /usr/bin/env node

'use strict';

var db = require('./models'),
    optimist = require('optimist'),
    _ = require('lodash'),
    path = require('path');

_.each(optimist.argv._, function (file) {
    var module = require(path.join(process.cwd(), file));
    module.seed(db.sequelize);
});

Users-seed-example.js

var _ = require('lodash'),
    fs = require('fs'),
    path = require('path'),
    csv = require('csv'),
    Sequelize = require('sequelize');

function addUser(db, filename, rolesByName, addresses) {
    var parser = csv.parse({}, function (err, data) {
        db.models.User.bulkCreate(
            _.map(data, function (record, index) {
                // Create user attributes object and return it
            })
        );
    fs.createReadStream(filename).pipe(parser);
}

var seed = function (db) {
    Sequelize.Promise.all([
        db.models.Role.findAll(),
        db.models.Address.findAll()
    ]).spread(function (roles, addresses) {
        var rolesByName = _.each(roles, function (role) {
            rolesByName[role.name.toLowerCase()] = role;
        });
        addUsers(db, path.join(__dirname, 'data', 'users.csv'), rolesByName, addresses);
    });
};

module.exports = {
    seed: seed
};

Contrived example of seeding Sequelize User objects from some CSV file where each row in the CSV file references some role (or more) using a role name, hence proprocessing the roles.

Thanks. I made this based on your code:

fs.createReadStream(path.join(__dirname, 'data', 'airports.csv')).pipe(csv.parse({}, function (err, data) {
        db.Airport.bulkCreate(
            _.map(data, function (record, index) {
                return {
                    'code': record[0],
                    'latitude': record[1],
                    'longitude': record[2]
                }
            })
        );
    }
));