drift-spark 0.3.0

Spark-native embedding lifecycle — produce, CDC refresh, migrate, audit.

drift-spark

Spark-native embedding lifecycle — produce, CDC refresh, model-migrate, audit.

pip install drift-spark · import drift · MIT

---

Status: pre-alpha (v0.0.1 placeholder).

Drift is a Python library that turns the standard 300-line PySpark embedding pipeline into three declarative commands:

drift embed --table my_catalog.docs --text-col body --sink qdrant://localhost:6333/docs
drift watch --table my_catalog.docs --text-col body --sink qdrant://localhost:6333/docs
drift status --sink qdrant://localhost:6333/docs

What it does:

• embed() — Spark-native embedding with dedup, batching, multi-model, Qdrant + pgvector sinks
• watch() — incremental CDC refresh via Delta Change Data Feed → only changed rows re-embedded
• migrate() — dual-write model migration with lineage tracking (v1.0); adapter projection (v2.0)
• Lineage ledger — per-embedding cost, source tracing, GDPR-delete proof (SQLite, queryable)

GitHub: https://github.com/aayush4vedi/drift-spark PyPI: https://pypi.org/project/drift-spark/0.0.1/