How would you implement direct-to-S3 file uploads from a client, avoiding routing large file uploads through your Node.js server entirely?

Rather than having the client upload a file to your Node.js server, which then re-uploads it to S3 (doubling bandwidth usage and tying up server resources for the duration of potentially large uploads), you generate a pre-signed S3 URL server-side (a temporary, cryptographically-signed URL granting time-limited permission to upload directly to a specific S3 location) and return it to the client, which then uploads the file directly to S3 using that URL, with your server never touching the actual file bytes at all.

const { S3Client, PutObjectCommand } = require('@aws-sdk/client-s3');
const { getSignedUrl } = require('@aws-sdk/s3-request-presigner');

app.post('/upload-url', async (req, res) => {
  const command = new PutObjectCommand({ Bucket: 'my-bucket', Key: `uploads/${crypto.randomUUID()}` });
  const url = await getSignedUrl(s3Client, command, { expiresIn: 300 });
  res.json({ uploadUrl: url });
});