Update: While ODM libraries like Mongoose make working with MongoDB easier, it's important to remember how the aggregation pipeline is used to efficiently query data in Mongo.
Mongoose is an object document modeling (ODM) layer that sits on top of Node's MongoDB driver. If your coming from SQL, it's similar to an ORM for a relational database. While it's not required to use Mongoose with the Mongo, here are four reasons why using Mongoose with MongoDB is generally a good idea.
MongoDB is a denormalized NoSQL database. This makes it inherently schema-less as documents have varying sets of fields with different data types. While this provides your data model with flexibility as it evolves over time, it can be difficult to cope with coming from a SQL background. Mongoose defines a schema for your data models so your documents follow a specific structure with pre-defined data types.
Mongoose has built in validation for schema definitions. This saves you from writing a bunch of validation code that you have to otherwise write with the MongoDB driver. By simply including things like required:true in your schema definitions, Mongoose provides out-of-the-box validations for your collections (including data types).
Mongoose provides optional pre and post save operations for data models. This makes it easy to define hooks and custom functionality on successful reads/writes etc. You can also define custom methods that act on a particular instance (or document). While you can achieve similar functionality with the native MongoDB driver, Mongoose makes it easier to define and organize such methods within your schema definition.
Mongoose makes returning updated documents or query results easier. A prime example can be found with update queries. While the native driver returns an object with a success flag and the number of documents modified, Mongoose returns the updated object itself so you can easily work with the results.
There are few reasons not to use Mongoose with MongoDB (especially if you are just getting started). For more advanced queries, it can be argued that Mongoose makes things more difficult and can slow performance. Advocates of the native MongoDB driver also argue that bringing ODM to a denormalized design entirely defeats the purpose of a NoSQL database.
Despite the arguments against using Mongoose, it remains one of the most popular ODM tools for Mongo. If you are coming from a SQL background then using Mongoose will make the transition into a NoSQL environment much easier. It will also save you time writing your own validations and instance methods and is highly recommended for smaller DBs and basic Mongo operations.