Data collection

Event bus & workers

Redis Streams as ephemeral transport between services, with a standard worker framework for background jobs.

Redis is an ephemeral coordination layer — event streams, job queues, rate limiting, distributed locks, short-term cache, deduplication and workflow signals. It is never a system of record.

The streams

market.events
listing.events
onchain.events
social.events
research.requests
feature.requests
scoring.requests
position.evaluation
execution.requests
execution.results
risk.events
system.alerts

The event envelope

Every event carries a common envelope so any service can consume it:

redis event

event_id · event_type · source_service · entity_type · entity_id
occurred_at · published_at · schema_version · correlation_id
payload · retry_count

Workers

Option	When
Dramatiq	Preferred — a lightweight Python worker architecture
Celery	When advanced scheduling, routing and mature tooling are needed
Custom async worker	Only for latency-sensitive streams

Redis is not permanent storage

Durable truth lives in Convex, SQLite and Parquet. Redis carries events in flight — replayable and disposable.