Skip to content
Eremie HQ

Siphon

A high-throughput bookmaker data ingestor with a clean star schema.

โ— LIVE2026

An async orchestrator that ingests fixtures, full odds time-series, and period-level results across multiple sources and sports, canonicalises the same real-world event across providers, and lands it in a normalised Postgres store for downstream models.

Stack

PythonasyncioPostgreSQLFastAPIstar schema

Engineering highlights

  • Odds stored as deltas (one row per price change) in a normalised star schema โ€” full line-movement history at a fraction of the size.
  • Deterministic cross-source canonicalisation so one real-world fixture is a single row, linked to its league.
  • Pure-HTTP adapters where adding a source is one adapter + a registry line; every quote captured, mapped or not.
  • Read-only FastAPI surface returning a typed response envelope for dashboards and models.
Eremie Gillowei ยท Preston, UK
eremiehq.com