Build data workflows that flow to production.
Phlo is a modular Python framework for lakehouse applications. Add packages for orchestration, storage, ingestion, quality, observability, and UI extensions, then run the platform from one project.
# choose the lakehouse services you need
$ uv pip install phlo[defaults]
$ phlo init my-lakehouse
$ phlo services start
dagster minio nessie trino
# materialize assets with checks and lineage
$ phlo materialize --select "dlt_glucose_entries+"
checks passed · lineage updated
$ phlo logs --asset glucose_entriesRun a lakehouse as one project.
Phlo wraps the moving parts of a lakehouse into a Python project: services, workflows, materializations, checks, lineage, and observability.
- 01
Bootstrap a project
Create the workspace, configuration, env files, and workflow folders Phlo expects.
- 02
Start the lakehouse stack
Bring up the services your project uses: orchestration, storage, catalog, query, and UI.
- 03
Materialize and inspect
Run assets, evaluate checks, then inspect lineage, logs, metrics, and service state.
Project
phlo.yaml, env, workflows
Services
Dagster, MinIO, Nessie, Trino
Pipelines
DLT ingestion, dbt transforms
Quality
schemas and checks on runs
Catalog
branches, tables, lineage
Observe
Observatory, logs, metrics
The point is not the plumbing. The point is a repeatable run with data, checks, lineage, and service state in one place.
Move from source data to governed assets.
Phlo gives the common lakehouse path a project shape: ingest data, merge it into tables, transform it, validate the run, and make the result observable.
# create the project and bring up the stack
$ uv pip install phlo[defaults]
$ phlo init glucose-lakehouse
$ phlo services start
dagster minio nessie trino observatory
# run assets and inspect what happened
$ phlo materialize --select "dlt_glucose_entries+"
materialized assets, checks evaluated
$ phlo logs --asset glucose_entries --since 1h
$ phlo lineage show glucose_entries- 01
Ingest data
Fetch from APIs, files, and databases, then stage data before it reaches tables.
- 02
Store and branch
Write to Iceberg-backed storage with catalog branches for reviewable lakehouse changes.
- 03
Transform and check
Run dbt or Python workflows and evaluate quality checks during materialization.
- 04
Observe the run
Use Observatory, logs, lineage, and metrics to understand pipeline health.
One project surface
Configuration, workflows, services, schemas, and transforms live in the same project.
Open lakehouse stack
Phlo coordinates tools like Dagster, MinIO, Nessie, Trino, dbt, and Observatory.
Run visibility
Materializations carry checks, logs, lineage, metrics, and enough context to debug.
Designed for open lakehouse stacks
Phlo's core stays small. Packages bring the lakehouse services and integrations, while open formats keep the data portable.
CLI · services · checks · materialization
Services
Postgres, MinIO, Nessie, Trino, Dagster, and optional stack packages.
Sources
Connector packages bring APIs, databases, files, and domain inputs into the lakehouse.
Quality
Checks run with materialized assets and pipeline executions.
Transformations
dbt and Python packages turn raw tables into modeled assets.
Cataloging
Nessie and OpenMetadata packages track branches, schema, and lineage.
Observability
Prometheus, Grafana, Loki, logs, metrics, and alerts plug in as packages.
Observatory
UI extension packages add routes, slots, settings, and dashboards.
Runtime
Adapters translate package contributions into the active workflow engine.
Guides that explain the shape of Phlo
Start with the package model that users install and operate, then go deeper only when you are authoring an integration package.
Getting started
Install phlo, add the packages you need, start services, and materialize the first pipeline.
Read guideGuide 02Plugin system
How to build custom sources, quality checks, transformations, and services for Phlo.
Read guideGuide 03Custom packages
How teams package domain assets, resources, services, and reusable integration logic.
Read guideGuide 04Observatory extensions
How packages add routes, dashboard slots, settings, and UI modules to Observatory.
Read guideShip governed data workflows with confidence.
Install core Phlo, add the packages your stack needs, start the services, and materialize assets with checks attached to the run.
$ uv pip install phlo $ uv add phlo-dagster phlo-nessie phlo-minio phlo-trino $ uv add phlo-dbt phlo-quality phlo-observatory $ phlo init my-lakehouse $ phlo services start services ready: dagster, minio, nessie, trino $ phlo materialize --select "dlt_glucose_entries+" materialized assets with package-provided checks