A software engineer who tries to document problems I run into and what I learn from them (hopefully it is useful to you too).
Building a Low-Volume Batch Worker Without a Database: Using S3 as Task Storage
ETags for compare-and-set, an inverted ULID for newest-first ordering, a zero-byte marker as a polling index — turns out that’s enough for a real task queue, no database or queue service needed.