jtagliabuetooso

5 ms·

31.
by jtagliabuetooso 2y ago
Thanks a lot for the feedback, point taken! Wrt deployment: the system has a control plane (on Bauplan AWS, never see any data, just auth and metadata), and data planes for customers (single tenant, private link, Soc2 compliant and all that
32.
by jtagliabuetooso 2y ago
Thanks @sbpayne <3
33.
by jtagliabuetooso 2y ago
In a nutshell, data and AI workloads require fast re-building and vertical scaling: 1) you should not need to redeploy a Lambda if you you're running January and February vs only January now. In the same vein, you should not need to re
34.
by jtagliabuetooso 2y ago
Looking to get feedback for a code-first platform for data: instead of custom frameworks, GUIs, notebooks on a chron, bauplan runs SQL / Python functions from your IDE, in the cloud, backed by your object storage. Everything is version
35.
by jtagliabuetooso 2y ago
+1 to kuzu: kuzu runs as ephemeral memory graph at every request in Bauplan as of today - which means 10k times a day at least ;-) We shared our use case publicly, with a bit of details on performance gains and simplicity: https:/&#x2
36.
by jtagliabuetooso 3y ago
Quack-reduce author here. I think the most interesting thing about duckDB (for us) is that you can "take out the DB" from duckDB and use the built-in engine over a (stateless, ephemeral) table you got from wherever. We didn't
37.
by jtagliabuetooso 4y ago
As much as self-citation is bad ;-), I want to share the live coding session we had last month: how to build a recommender system pipeline entirely in the browser ( https://www.youtube.com/watch?v=PkKs6oWREr4 ) through the Ou