creativedutchmen

Data Scientist at Jedlix

[creativedutchmen @creativedutch]

My entire twitter feed right now

[creativedutchmen @creativedutch]

I cannot overstate how much better this makes Amazon. This is soooooo good, and it’s kind of insane they let the product rot this far in the first place

[Josh Pigford @Shpigford]

Knockoff is now live!

Filter out the knockoff crap brands on Amazon.

Sorry to brands like WNPETHOME, EHEYCIGA, YXYL, LU&MN, JOYIN, TOMY, GODONLIF, YOOJEE, LINGTENG, LANEIGE, VISCOO, BIODANCE, COOFANDY, BALENNZ, TOSY and LUENX.

https://t.co/9mLk0EAsfG

[creativedutchmen @creativedutch]

@runaway_vol Certified banger

[creativedutchmen @creativedutch]

@markoilico Yes please

[creativedutchmen @creativedutch]

@iannuttall It’s pretty great as is, and could be totally amazing with a few additions, like git worktree support, and quick access to run shell commands.

[creativedutchmen @creativedutch]

Don’t know why, but gpt5 is really getting on my nerves… unhelpful, even worse “you are spot on in your analysis”… wasn’t this model supposed to be better?

[creativedutchmen @creativedutch]

@ryanwinchester Unless you’re using sqlserver, because those geniuses decided to use a different sorting method for whatever reason

[creativedutchmen @creativedutch]

@dhh Big servers are cheap, too. We’re running our sentry instance on a single €200/mo hetzner machine. That gets you 48 true cores, 256GB ram, ~7TB raided ssd storage…

Who needs autoscaling when resources are this cheap?

[creativedutchmen @creativedutch]

@vsbuffalo No need for poetry, https://t.co/LdhsvbQWQt compiled to requirements.txt is all you need :) compile with uv, install with uv. Fast, painless

[creativedutchmen @creativedutch]

@jimseven @Keen_Coffee this has your name written all over it :)

[creativedutchmen @creativedutch]

@finbarrtimbers @jgschraiber Karpathy in his transformer from scratch video said something along those lines iirc

[creativedutchmen @creativedutch]

@JonathanNFL_ @matsonj A great way to waste a holiday is what it is…

[creativedutchmen @creativedutch]

@rlmcelreath Looks good! That crust would benefit from a little more oven time though. IMO taking it out right before “oh crap it’s burned” delivers the best results

[creativedutchmen @creativedutch]

@leftofshoes @kemery7 @dieworkwear @FinzZombie Kind of like the look, any hints on search terms or things to look for?

[creativedutchmen @creativedutch]

@RitchieVink What dashboard is this?

[creativedutchmen @creativedutch]

@laurenbalik @ergestx Yeah in my case we have a bunch of “not so standard” data, so there’s going to be a bunch of enrichment in the warehouse itself no matter what.

We’re planning a nice rework of most of our db, so this is good time as any to build on these ideas.

Are there any good posts to read?

[creativedutchmen @creativedutch]

@laurenbalik @ergestx Cool! One thing I’d really like about this idea is that it can capture changing state with ease - if you add the data early, that is.

I’m still struggling to put this in practice - do you suggest to join the dim tables early, late, or do without them completely?

[creativedutchmen @creativedutch]

@laurenbalik @ergestx Hearing this sentiment more and more, what’s your preferred solution? No more fact dim tables, but all in one big, wide table?

[creativedutchmen @creativedutch]

@matsonj @__AlexMonahan__ I guess my usecases involve too much centralization for this to really make sense, and something like clickhouse is probably a better fit. Thanks for the pointers to, super interesting to hear your angles and solutions!

[creativedutchmen @creativedutch]

@matsonj @__AlexMonahan__ Yeah, that’s super similar to the pattern I’m already using - note the timings going to absolute hell when the nr of files increases. 10+s for an operation on <1M rows is not great