Platform Eng
— Week of Mar 10

What shipped, what's moving, and one decision we need from the room before the Acme 2.4 cut.

Friday demo 6 slides ~4 min
Shipped this week

Three things out the door

  • Bulk task import
    CSV and JSON uploads now land straight into a board with column mapping — no more paste-and-pray.
    #4211
  • Webhook retries v2
    Exponential backoff with jitter; dead-letter queue surfaces in the workspace admin panel.
    #4188 #4203
  • Postgres 16 migration
    All read replicas cut over Tuesday night; zero customer-facing downtime, ~9% faster aggregate queries.
    #4179
In progress

Carrying into next week

  • Recurring tasks engine ~70%

    Scheduler and RRULE parsing are done; remaining work is the timezone edge cases and the "skip holidays" toggle.

  • Audit log export ~35%

    Streaming NDJSON endpoint is up behind a flag; still wiring the S3 destination picker and retention policy UI.

Metrics

Numbers we watch

API p95 latency
184ms
↓ 12% wk/wk
Background job error rate
0.21%
↓ 0.08pp
p95 latency, trailing 8 days — lower is better
Decision needed

One call to make

Do we ship recurring tasks behind a workspace flag in 2.4, or hold one more week for the timezone fixes?

Flagged rollout gets it to design partners Friday but means two code paths for ~2 weeks. Holding keeps a single path but slips the partner promise.

A — Flag it, ship Friday B — Hold for 2.5
Next week

On deck

  • Finish recurring tasks — timezone matrix tests, then dogfood on the internal ops board.
  • Audit log export to private beta — three workspaces lined up.
  • Start scoping rate-limit headers for the public API; RFC draft by Thursday.
Questions → drop them in the platform channel or grab anyone after standup.
1 / 6