Sure, but there are tradeoffs. One option is to just concatenate all the drives together, either with mdadm (grouping the drives into a RAID-0 array) or with LVM (adding them all to a single volume group). This lets you add drives very easily and efficiently, but has no redundancy. When (not if) one drive fails, your entire FS is toast.
If you care about your data, then you'll want to use RAID-6 for redundancy. But adding a drive to a RAID-6 array is very time-consuming, because all of your data has to be re-shuffled onto a new drive. And you don't want to do this without a backup -- although the reshape process is designed to be safely interruptible, it's not bug-free. (I speak from experience, having nearly lost a few TB this way not too long ago.)
You can compromise by grouping your drives into multiple RAID-6 arrays. For example, instead of using 58 data drives + 2 parity drives, you could use 6 arrays of (8 data + 2 parity). This gives you much less usable space (80% as opposed to 97%) but it allows you to add 10 drives at a time without having to re-shuffle all your data. And it means you can recover from a drive failure much more quickly.
Of course, you can also just treat the drives as separate block devices, and provide a single redundant filesystem at a higher level, using something like Ceph. I would assume Backblaze is doing something along these lines.