iiinova — where your digital self keeps evolving

Emily Riederer @emilyriederer.bsky.social · 11m

Sounds wonderful and really resonates! Will there be a recording?

Emily Riederer @emilyriederer.bsky.social · 8h

Absolutely bonkers, IMHO, that BigQuery table names are case sensitive by default O/C, everything uncontroversially is in R and python, but I feel that it's so anathema to the rest of SQL Just fixed a gnarly bug with `is_case_insensitive = FALSE` (after staring at the double neg for a bit) 58/

020

Emily Riederer @emilyriederer.bsky.social · 20h

I previously used Connect! (That team still does; I just changed roles where it became less relevant) Happy to chat about it

000

Emily Riederer @emilyriederer.bsky.social · 20h

Bound to happen but you hate to see it: technical diagrams in GCP docs with the AI slop aesthetic. Historically, their dev docs have been *very* nice (much better than AWS imho) So compact, so saturated.. 56/

110

Reposted by Emily Riederer

Noam Ross @noamross.net · 07/10/2026

I'm speaking at the NYC Statistical Computing Meetup about @grantwitness.org: how we use modern #rstats stack to produce trustworthy, reproducible, rapid data that powers activism, litigation, and journalism. Join us in-person @ NYU or online. Mon, Oct 19 at 7PM! www.meetup.com/nyhackr/even... 🧪

meetup.com

Grant Witness: An R-Powered Research and Advocacy Enterprise, Mon, Oct 19, 2026, 7:00 PM | Meetup

**[External registration](https://nyhackr.org/?utm_source=nyhackr)** required at **[nyhackr](https://nyhackr.org/?utm_source=nyhackr)**. This month we have Noam Ross retur

0125

Emily Riederer @emilyriederer.bsky.social · 08/10/2026

I regret to inform me they want more changes to this guy One day I will find time to properly tear it down and build it back up bsky.app/profile/emil... 55/

130

Emily Riederer @emilyriederer.bsky.social · 08/10/2026

In an icebreaker a few weeks ago, our team was sharing pet peeves and an early-20s said API 429 errors. Not API errors but specifically status code 429 (too many requests) and I'm still thinking about it If that somehow isn't a metaphor for early 20s exuberance.. your 30s are all 500s

0142

Emily Riederer @emilyriederer.bsky.social · 07/10/2026

Reminding myself about this "get stuff done" piece as I resist the urge to refactor Volunteering is great complement to day job because risk sloping between 'right' and 'fast' is quite different. Good to practice slow and fast twitch muscles bsky.app/profile/emil... 54/

130

Emily Riederer @emilyriederer.bsky.social · 07/10/2026

MailChimp IMHO has always struck me as a product with so little SWE sensibility. From the janky email templates lacking any sort of control flow (it makes you **miss** Jinja!) and apparently all the way to the API Gotta pass text blobs to API instead of an ID like a sane person 53/

150

Emily Riederer @emilyriederer.bsky.social · 06/10/2026

Well if I ever need to remember what to write on a post it later, I’ll consider a sketchpad!

000

Emily Riederer @emilyriederer.bsky.social · 06/10/2026

Sometimes during the day I make a note on my PostIt note to-do list that is just to remind me to add something to my running Google Doc to-do list. This is either very organized or very chaotic and I’m not even sure which anymore

481

Emily Riederer @emilyriederer.bsky.social · 06/10/2026

Perhaps with undecideds 50 isn’t as relevant of a benchmark?

010

Emily Riederer @emilyriederer.bsky.social · 06/10/2026

I guess it’s reasonable to have an “is this thing on” view, but imho maybe not the most salient failure mode to spotlight?

010

Emily Riederer @emilyriederer.bsky.social · 06/10/2026

I can confirm it is not errors in the log because these have plenty! In this case it’s the Google Cloud Function confirming the endpoint deployed correctly so it is callable — not the output of any of the umpteenth times it may have failed when called

110

Emily Riederer @emilyriederer.bsky.social · 06/10/2026

This job evidently has too few checks around actual failure modes yet copious error handling for env vars that will only ever be set once by the primary dev in a container they configured. It's not an input! 52/

110

Emily Riederer @emilyriederer.bsky.social · 06/10/2026

Speaking of those illusive green checks, this one job posted over 1500 failures in the last week (ngl tho, this little sparklines-ish table in nice) bsky.app/profile/emil... 51/

220

Emily Riederer @emilyriederer.bsky.social · 06/10/2026

IAM roles, man. Bane of my existence Funny example of "What is a corner case to test?" Job fails. Need Object Creator role. Done. Works. Run two. Job fails. Need higher access to delete. Done. Works. The base case "tested under same environment" keeps changing. 50/

100

Emily Riederer @emilyriederer.bsky.social · 06/10/2026

Another thing that helped was being more memory conscious w polars df.to_dicts() to df.iter_rows(named=True) to use DataFrame rows as a generator! 49/

120

Emily Riederer @emilyriederer.bsky.social · 06/10/2026

This makes me think about the missing 102 knowledge So many 101 tutorials, e.g. API calls. But I just swapped requests -> requests.Sessions and got a job to run in <5 min that was dying after 30+ Not relevant when not running at scale, but not a well known option then when trying to scale 48/

130

Emily Riederer @emilyriederer.bsky.social · 06/10/2026

A charming little thing about Cloud Run functions is how the little green checks mean succcessfully *deployed* while the logs themselves may be screaming out in pain and utterly failing Optimism 47/

250

Emily Riederer @emilyriederer.bsky.social · 04/10/2026

And I fear that we have reached the "cleaning up some AI slop" stage. Never seen emojiis in error logs before (And of course the print() for logging does not make log inspection any easier, either..) 46/

130

Emily Riederer @emilyriederer.bsky.social · 04/10/2026

I’m not sure if I’ve complained about this one yet but with cloud stuff man does it pay to be strategic about your regions Different datasets stored across diff regions. BigQuery has made this better but cross region is still super slow. A little planning goes a long ways 45/

110

Emily Riederer @emilyriederer.bsky.social · 02/10/2026

Yeah it’s so good at large data no one talks about how good it is at all! But so nice to stop dealing w corrupted CSVs

020

Emily Riederer @emilyriederer.bsky.social · 02/10/2026

Two corollaries tonight: - a log no one looks at is not observability - creation of a log should not require code as long and complex as the job itself bsky.app/profile/emil... 44/

241

Emily Riederer @emilyriederer.bsky.social · 02/10/2026

The eternal question in data is "what is the denominator" and man does Gemini make silly choices. Asked how common an injury is, one might assume I meant *in people* but am given two rates: - % of injury among all hamstring injuries - % of injury among those who have half of the injury 🤷‍♀️

1141

Emily Riederer @emilyriederer.bsky.social · 02/10/2026

Blows my mind that in 2026 Parquet is not yet the absolute default for data sharing A few years back there was a lovely talk at @rohanalexander.bsky.social 's Toronto Data Workshop by some students who articulated one of my soapboxes far better than I ever have: www.youtube.com/watch?v=Evcn...

youtube.com

Exploring Alternatives to REST for Accessing Public Data Sets

YouTube video by Rohan Alexander

3273

Emily Riederer @emilyriederer.bsky.social · 02/10/2026

Another little thing I think about a lot is how tools can seem quite silly until you have the exact need for them. You just can't judge. Long ago, I just could NOT get why you'd need Terraform or Pulumi. But holy cow, once you've felt the pain of clicking through a bajillion cloud consoles.. 43/

140

Emily Riederer @emilyriederer.bsky.social · 02/10/2026

I want to go to those sessions and read those books!

000

Emily Riederer @emilyriederer.bsky.social · 02/10/2026

Same thing with so many BI tools. This is why I'm grateful for totally open source options like Shiny / R Markdown. I've blogged about how to paywall Rmd on Netlify and now interested to look at Posit Connect Cloud for this group (3/3) 42/

150

Emily Riederer @emilyriederer.bsky.social · 02/10/2026

Right now I'm scheduling a few things with GitHub Actions* (*feels icky to me when its for CI but its cheap and accessible) but I worry a bit about it becoming inscrutable for the next person (2/3) 41/

150

Emily Riederer @emilyriederer.bsky.social · 02/10/2026

Another soapbox this reminds me: pricing models for data tools are SO hostile to "data teams of one" / non-profits. I think this team could learn a batteries-include tool like dbt Cloud (I prefer CLI but doubt they want local set up) but pricing makes no sense outside of enterprise (1/3) 40/

140

Emily Riederer @emilyriederer.bsky.social · 02/10/2026

Google Cloud will nearly force you to put dashes in project names (underscores not allowed) only to then force you to `backtick` any said use of project name in queries. WHY. 39/

140

Emily Riederer @emilyriederer.bsky.social · 01/10/2026

It really does! I'd love to see more public "reference architecture" type projects with these tools!

000

Emily Riederer @emilyriederer.bsky.social · 01/10/2026

I also don't want to brag, but it occurred to me this morning this delightful group has not said AI to me once 🙏 38/

170

Emily Riederer @emilyriederer.bsky.social · 01/10/2026

This but for all ideas. Shocking how many can fit in a 300-char post (or a pumpkin!) if you put your mind to the key idea.

020

Reposted by Emily Riederer

Eric Columbus @ericcolumbus.bsky.social · 01/10/2026

In all seriousness I wish every professor on here would post their reading lists each semester

2041436

Emily Riederer @emilyriederer.bsky.social · 01/10/2026

The cool thing is this group really cares about getting their data right. Had a (genuinely; not /s) fun convo tonight about how hard counting is. The perennial classic case of numbers from diff systems disagreeing without being wrong. Bcs data only reflects its source not a deeper reality 37/

160

Emily Riederer @emilyriederer.bsky.social · 01/10/2026

We have our first Excel bug! Got some spreadsheet data from a vol to load via an API. One of required fields is a hexadecimal hash. Excel (or Sheets?) click and drag started incrementing it instead of copying it down the column🤦‍♀️ Good reminder that our tools started to hallucinate before AI 36/

160

Emily Riederer @emilyriederer.bsky.social · 01/10/2026

If you automated something but do not have alerts to tell you when it errored out, no you didn't 35/

480

Emily Riederer @emilyriederer.bsky.social · 30/09/2026

Apparently I talk about knowing “what good looks like” too much because I started to say this today and my team finished my sentence 😳 That said, so glad to have “grown up” when #rstats was all about talking about tacit knowledge and workflows and the joy of setting things up nicely

3141

Reposted by Emily Riederer

David M. Perry @lollardfish.bsky.social · 29/09/2026

Bluesky, the app for books

643678

Emily Riederer @emilyriederer.bsky.social · 29/09/2026

I've been doing this long enough I need to start calling out myself too. Just caught myself massively overengineering a query. All sort of complexity to avoid a `cross join` which makes my inner DBA scream. But realistically, table is never big enough to be a problem. Those exist for a reason. 34/

151

Emily Riederer @emilyriederer.bsky.social · 29/09/2026

Pretty elegant given what you were working from! If I'd read the code first, I could not have fathomed with the comment "votes are every other key" could possibly mean

110

Emily Riederer @emilyriederer.bsky.social · 29/09/2026

I've never liked the name, but the open-close principle is hyper relevant to the issues here. Or more simply "require changes to as few and least important scripts as possible" So much business logic deep in the jaws of heavy lifting code 33/

140

Emily Riederer @emilyriederer.bsky.social · 29/09/2026

I'm sitting here alone but literally when I opened this said omg and started laughing to no one in particular 😆 Thank you -- given that I *don't* have to work with it, this file is a gift I'm now fascinated what horrific source system make this the most intuitive way for someone to publish...

110

Emily Riederer @emilyriederer.bsky.social · 29/09/2026

Also, love the idea of hidden open APIs. That had never even occurred to me until I heard @lucystats.bsky.social talk about the Disney one. Will check out your Lab episode!

120