Implementing joins on multiple collections - node.js

Question

I am trying to create a social networking application which can have connect (followers, following), posts, comments, likes, shares, etc. This is a MVP project, but still i wanted to explore mongoDB for this use case. I am having some doubt regarding the performance of this application.
I have three collection:
Posts: This is where a new post shall be added. This collection contains all the details related to a post.
Schema:
const postSchema = new mongoose.Schema({
user_id: String,
title: String,
text: String,
type: { type: String, enum: ["music", "movie", "tv"] },
mediaUrl: String,
thumbnailUrl: String,
accessType: { type: Number, enum: [1, 2, 3] },
tags: [String],
like_count: Number,
comment_count: Number,
share_count: Number,
insertDate: {
type: Date,
default: () => {
return new Date();
}
}
});
Feeds: This collection just add a metadata of user and post including tags. This i intend to use to get the relevant feed for a particular user.
Schema:
const feedSchema = new mongoose.Schema({
post_id: String,
user_id: String,
isTag: Boolean,
isPublic: Boolean,
insertDate: {
type: Date,
default: () => {
return new Date();
}
},
modifiedDate: {
type: Date,
default: () => {
return new Date();
}
}
});
Connects: This collection is for the relationship of users.
Schema:
const connectSchema = new mongoose.Schema({
followed_by: String,
user_id: String,
insertDate: {
type: Date,
default: () => {
return new Date();
}
}
});
My approach was to first find the posts from feeds collection basis users whom I am following, then fetching the posts from post collection.
Here goes my attempted query:
db.connects.aggregate([
{ $match: { followed_by: "5cbefd61d3b53a4aaa9a2b16" } },
{
$lookup: {
from: "feeds",
let: { user_id: "$user_id" },
pipeline: [{ $match: { $expr: { $or: [{ $eq: ["$user_id", "$$user_id"] }, { isPublic: true }] } } }],
as: "feedData"
}
},
{ $unwind: "$feedData" },
{ $replaceRoot: { newRoot: "$feedData" } },
{ $group: { _id: { post_id: { $toObjectId: "$post_id" }, modifiedDate: { $toLong: "$modifiedDate" } } } },
{ $replaceRoot: { newRoot: "$_id" } },
{ $sort: { modifiedDate: -1 } },
{ $skip: 0 },
{ $limit: 10 },
{
$lookup: { from: "posts", localField: "post_id", foreignField: "_id", as: "postData" }
},
{ $unwind: "$postData" },
{ $replaceRoot: { newRoot: "$postData" } },
{ $addFields: { userId: { $toObjectId: "$user_id" } } },
{
$lookup: { from: "users", localField: "userId", foreignField: "_id", as: "userData" }
},
{
$project: {
post_id: "$_id",
user_id: 1,
title: 1,
text: 1,
typeaccessType: 1,
mediaUrl: 1,
thumbnailUrl: 1,
insertDate: { $toLong: "$insertDate" },
like_count: 1,
comment_count: 1,
share_count: 1,
user_email: { $arrayElemAt: ["$userData.email", 0] },
user_profile_pic: { $arrayElemAt: ["$userData.profile_pic", 0] },
username: { $arrayElemAt: ["$userData.firstname", 0] }
}
}
]).pretty();
Please share your feedback on:
Which index should I use to boost up the performance?
Is there is a better way of doing the same in mongoDB. Also if any part of the query can be optimised?