Made a quick comparison between two license plate blurring models: NVIDIA LPDNet and Meta EgoBlur. Which one is the better in terms of accuracy and performance? https://t.co/EpYeIPgo96
@kellabyte their helm chart has a nice sidecar which deploys dashboards from configmaps, or just add stuff to dashboardProviders in values.yaml.
tl;dr it's funky if you deploy dashboards from helm
@GergelyOrosz@deejayyhu nekem is csak ez jött be, ha nem is pénz miatt, hanem mert szégyen lemondani. Ha mázlid van, akkor a Stockholm szindróma miatt talán még a mozgást is megszereted
Is @dagsterio cool? Hell yes, at least for data pipelines. Why I prefer its new software-defined assets concept, what value it adds to the traditional tasks/dependency graph model, and how easy to integrate with common tools like @AirbyteHQ and @getdbt: https://t.co/DCpuvLOPVU
@aleksizy@AirbyteHQ@wrike@meltanodata I tried stitch back then a while but run into some issues and give up. maybe they evolved since then, but I would not recommend it
I tried how it feels to build a source connector for @AirbyteHQ. It felt so good that I wrote a post on how I made a source connector for @wrike:
https://t.co/Q7F3aRJxFO
@aleksizy@AirbyteHQ@wrike wrt singer based products I heard many good feedbacks about @meltanodata , but never tried myself. The MIT license and the number of connectors are a great plus. However, it is too python centric for me, you simply cannot write fast code in python no matter what people say.