Latest

How I turned time-grouped search logs into transaction baskets, mined association rules with support, confidence, and lift, and validated a versioned recommendation batch before indexing it.

How I used a Redis List as a bounded replay window for reconnecting conversations, handled reverse ordering and memory limits, and separated a fast cache from durable message history.

How I implemented PostgreSQL outbox polling with SKIP LOCKED, explicit message states, and bounded retries, and where hard-crash recovery needs an additional policy.

How I implemented a Kafka notification consumer with CAS state transitions, separate data and infrastructure error paths, and deliberate acknowledgment ordering.

Why a notification pipeline used one PostgreSQL transaction for business data and event intent, then separate Kafka publishing and consuming workers—and what that structure does not guarantee.