The Customer

LIQID is a fast-growing fintech company based in Germany, offering digital wealth and portfolio management, access to alternative investments, and platform and advisory tools for private and professional investors. With over €3 billion in assets under management and more than 8,000 clients, LIQID depends on reliable, timely data to support sales performance tracking, marketing ROI analysis, and operational decision-making.

As the business scaled, LIQID needed a modern data platform that could keep pace – without requiring a large in-house data engineering team.

The Challenge

LIQID’s data orchestration layer was built on Google Cloud Composer and maintained by a small team. Over time, the setup became increasingly difficult to operate and expensive relative to actual usage.

Key challenges included:

• High fixed costs: Composer incurred disproportionate fixed costs despite pipelines running only once or twice per day.
• Operational rigidity: Composer supported only a single worker type, forcing all tasks to run on oversized, high-memory, and multicore nodes.
• Maintenance risk: Custom, repetitive code created a “black box” environment that teams were afraid to restart, resize, or upgrade.
• Limited trust in data: Business users lacked confidence in data quality and transformations.
• Growing backlog: Adding new data sources required significant manual effort and engineering time.

With a one-person engineering team responsible for orchestration, LIQID needed a solution that reduced operational burden while improving transparency, scalability, and cost control.

The Solution

LIQID partnered with Untitled Data, a data engineering consultancy specializing in modern data platforms, to redesign its orchestration layer. Together, they migrated LIQID from Google Cloud Composer to Astronomer, establishing a production-grade data orchestration platform built for scale, reliability, and developer efficiency.

Rather than simply replacing one scheduler with another, LIQID used Astronomer to standardize how data pipelines are built, deployed, and operated – creating a foundation that supports analytics today and more advanced use cases in the future.

Key elements of the solution included:

Standardized Ingestion

• Replaced brittle custom scripts with dlt for highly customized sources and Airbyte for standard SaaS connectors.
• Centralized orchestration of dlt, Airbyte, dbt Cloud, and Elementary within Astronomer, providing a single control plane for the data stack.

Granular Infrastructure

• Introduced infrastructure as code for repeatable, version-controlled platform setup.
• Shifted 80% of tasks to low-cost “micro-workers,” reserving high-memory nodes only for workloads that required them.

Cost-Aware Operations

• Implemented scheduled hibernation to automatically scale down the platform during nights and weekends.
• Migrated large Salesforce tables to incremental processing, reducing pipeline runtime from over an hour to under a minute.

By adopting Astronomer as a managed data orchestration platform, LIQID eliminated the operational overhead of running Airflow while gaining enterprise-grade reliability, observability, and extensibility – without increasing team size.

The Results

LIQID achieved immediate and measurable improvements after migrating to Astronomer:

• 63% reduction in orchestration costs by aligning platform spend with actual usage
• Up to 98% faster pipeline runtimes, with hour-long pipelines reduced to under a minute
• 2× higher compute throughput while operating Airflow with a lean 1-2 person team
• Improved reliability and confidence, eliminating fear around restarts, upgrades, and platform changes

“Astronomer gave us confidence in our orchestration layer. We no longer worry about upgrades or restarts, and our data pipelines just run.”

Jeffrey Lek
Head of Data, LIQID

Partner Perspective

Untitled Data served as LIQID’s implementation partner, bringing deep expertise in modern data platforms, pipelines-as-code, and cost-efficient cloud architectures. Working closely with LIQID’s lean data team, Untitled Data led the migration away from Cloud Composer, redesigned ingestion patterns, and established best practices that made the platform easier to operate, extend, and trust. With deep familiarity with the unique data environments and regulatory requirements across EMEA – including data sovereignty and financial data governance considerations – Untitled Data ensured the new platform aligned with regional compliance expectations from day one.

Beyond the initial migration, Untitled Data helped LIQID standardize how pipelines are built and maintained – reducing custom code, improving data quality, and ensuring the platform could be operated long-term without dedicated Airflow specialists.

“By combining Astronomer with standardized, code-first ingestion using dlt and Airbyte, we helped LIQID simplify operations, dramatically reduce costs, and create a data platform their small team can confidently run and evolve.”

Willi Müller
Co-Founder, Untitled Data

What’s Next

With a stable, cost-efficient orchestration platform in place, LIQID is focused on building on top of its modern data foundation.

Next initiatives include:

• Expanding automated sales and marketing performance reporting with higher data freshness and trust
• Accelerating onboarding of new internal and external data sources using standardized ingestion patterns
• Leveraging its reliable, well-orchestrated pipelines as a foundation for advanced analytics and future AI- and machine learning-driven use cases

With Astronomer and Untitled Data, LIQID now has a scalable, AI-ready data platform that can evolve alongside the business – without increasing operational complexity.