XBE · Software Engineer Core infrastructure: Postgres, scheduled workloads, archival and production observability.
- Query and index work on a large Postgres estate: finding why the planner was not using an index and fixing it in the query rather than by adding another, and auditing usage across primary and replica before anything was dropped.
- Designed and built a forward proxy in Go over a private mesh network — TCP and UDP relay, SOCKS5 and HTTP CONNECT — replacing per-container VPN daemons and command wrappers.
- Migrated a large fleet of scheduled jobs between hosting platforms as generated infrastructure-as-code, in waves, re-sizing after diagnosing memory limits on the first.
- Built and operated an archival pipeline moving cold partitions off an OLTP primary into columnar files on object storage, then hardened it so a stalled worker failed loudly and archival could not take an exclusive lock on a live table.
- Instrumented web and worker processes so container memory kills became attributable, and put database capacity on a daily report: disk runway, invalid indexes, unused candidates.