architectureIntermediate
Why we run DuckDB behind Parquet in the F1 pipeline
A one-file warehouse has real tradeoffs. Here's how we split writes to Parquet and kept DuckDB read-only for serving.
DuckDBParquetdbt
@marco
8 min read·3d ago
// blog
Engineering writing, not motivational content.
Architecture breakdowns, postmortems, debugging stories, workshop recaps.
postmortemAdvanced
Rebalancing storm: what a bad Kafka assignor cost us
A 20-minute lag spike, one flipped consumer flag, and a lesson in how partition assignment strategy is not a detail.
KafkaConsumer Groups
@nadia
12 min read·1w ago
debuggingBeginner
Chasing ODBC Driver Error 126 across three OSes
A story about dependency chains, VC++ runtimes, and why 'works on my machine' is a valid bug report.
ODBCWindowsLinux
@leo
6 min read·2w ago
communityBeginner
First-PR spotlight: @jules ships the Iceberg sandbox
From typo fix to project donation in six months. How the sandbox lowered the barrier to donating real systems.
IcebergCommunity
@team
4 min read·3w ago