havn -- Data in safe waters¶

havn is a self-hosted data platform -- a lightweight, Nordic alternative to Databricks and Snowflake. It uses DuckDB for OLAP analytics, plain SQL for transforms, and Python for ingest/export scripts. All data lives in a single warehouse.duckdb file. No data leaves your machine.

Quick Start¶

pip install havn
havn init my-project
cd my-project
havn jobs run full-refresh
havn serve

Or with Docker:

docker run -v "$(pwd)":/project -p 3000:3000 ghcr.io/chraltro/havn

Why havn?¶

	havn	Databricks	dbt + Airflow	Snowflake
Runs on	Your machine	Cloud	Cloud / self-hosted	Cloud
Data stays	Local	Vendor cloud	Depends	Vendor cloud
Setup time	1 minute	Hours	Hours	Hours
Cost	Free	$$$	$ - $$$	$$$
SQL dialect	DuckDB (Postgres-compatible)	Spark SQL	Varies	Snowflake SQL
Templating	None -- plain SQL	Jinja	Jinja	None

Features¶

Transform Engine¶

Plain SQL with @config directives. Dependencies are auto-extracted from FROM and JOIN clauses, so the DAG builds itself. SHA256 content hashing means only changed models rebuild. No Jinja, no templating, just SQL.

Data Quality¶

Inline assertions (@assert row_count > 0, @assert unique(id)), YAML contracts, column profiling, and freshness monitoring.

Connectors¶

Pre-built connectors for PostgreSQL, MySQL, Stripe, Shopify, Google Sheets, REST APIs, S3/GCS, CSV, and webhooks.

Web UI¶

React SPA with Monaco code editor, interactive DAG visualization, table browser, query runner, and notebook support.

AI-Native¶

Ships with CLAUDE.md, .cursorrules, and Copilot instructions. Run havn context to generate an AI-ready project summary.

Single-File Database¶

Everything in one warehouse.duckdb file. Back up with cp. No servers, no clusters, no cloud accounts.

Architecture¶

Ingest Scripts (.py / .dpnb)
        |
        v
  +-----------+      +----------+      +----------+      +----------+
  |  landing  | ---> |  bronze  | ---> |  silver  | ---> |   gold   |
  |  (raw)    |      | (cleaned)|      | (business|      | (consume)|
  +-----------+      +----------+      +----------+      +----------+
        ^                                                       |
        |                                                       v
   External Sources                                      Export Scripts
   (APIs, DBs, files)                                    (reports, APIs)

landing -- Raw data ingested from external sources via Python scripts or notebooks
bronze -- Light cleanup: column renaming, type casting, deduplication
silver -- Business logic: joins, aggregations, calculations
gold -- Consumption-ready tables for dashboards, APIs, and reports

Documentation¶

Category	Pages
Getting Started	Getting Started, Configuration, Environments
Core Concepts	Transforms, Python Models, Pipelines, Seeds, Sources, Macros, Packages, Semantic Layer
Branches and CI	Branches -- a warehouse per git branch, data diff on pull requests
Data Integration	Connectors, CDC, Live Models
Data Quality	Quality, Unit Tests, Contracts, Lineage
Refactoring	Refactoring -- rename a column, find column references
Security	Auth, Masking, Governance -- row-level security, lineage-following masking, governed Python
Sharing and Reports	Dashboards and Sharing -- published dashboards, scheduled reports
AI	Ask -- ask questions of your metrics, verified agent changes; MCP Server
Advanced	Scheduler, Notebooks, Versioning, Performance, Telemetry
Reference	What havn Supports, CLI Reference, API Reference