{"id":50411800,"title":"Show HN: stardelt.io – a self-hosted lakehouse on k8s (pre-alpha)","points":1,"user":"schwarztrinker","time":1794131433,"time_ago":"8 hours ago","type":"link","content":"

I am pretty honest: there are a ton of vibecoded projects around, and I made another one... great huh?

I'm working as a data platform engineer in Germany. My day job is basically spinning up & running lakehouse platforms on US hyperscalers for SMB and some DAX companies. \nI mean, data engineers like to work on these platforms but since some time I really have ethical concerns about it - like: \n- vendor lock-in\n- high costs (because of cloud costs - and in most cases you are not able to connect your own metal if you have one)\n- non-EU vendors...\n- other open-source solutions have high license costs for LDAP \netc. pp.

That's why is started to build my own opinionated data lakehouse platform in my spare time (tbh I also started learning agentic engineering and what harnesses are..), only curated from open source projects already available.

The building blocks are pretty straightforward imo:\n- Iceberg tables on whatever S3 storage you already have\n- Trino for SQL and Spark Connect for jobs\n- Airflow for schedules, plus notebooks, Kafka streams and dashboards\n- One web UI and one API over all of it\n- stuff around like OIDC, secrets management etc.

It's fully pre-alpha. But I want to use it as an internal pitch project at my current employer to push agentic engineering & the idea of creating our own data platform for our non-hyperscaler customers.

Site: https://stardelt.io\nCode: https://github.com/stardelt

I'm happy to hear what you think about it :)

Cheers, Martin","url":"https://stardelt.io/","domain":"stardelt.io","comments":[],"comments_count":0}