Michael Mullarkey

Michael Mullarkey

@mcmullarkey.bsky.social

Clinical psych PhD | ML Engineer in sports analytics | he/him My newsletter on building cool data stuff: https://buttondown.com/datadash Personal website: mcmullarkey.github.io Opinions are my own and don't reflect the opinions of my employer

4,503 Followers 1,135 Following 894 Posts Joined May 2023
14 hours ago
Preview
GitHub - DrCatHicks/learning-opportunities: A Claude Code skill for deliberate skill development during AI-assisted coding A Claude Code skill for deliberate skill development during AI-assisted coding - DrCatHicks/learning-opportunities

The learning opportunities skill from @grimalkina.bsky.social is so good. Sooo helpful at my new job. For thoroughly understanding code before I modify it. Catching bugs others had missed because it helps me think through the code so thoroughly!

github.com/DrCatHicks/l...

15 2 1 1
2 days ago

Building stuff where I can't tell if I'm cooking or getting cooked

2 0 0 0
4 days ago

Hello this is also me

1 0 0 0
4 days ago

Nothing quite like people using stuff you helped build to learn!

5 0 1 0
4 days ago

I like this tweak a lot! @grimalkina.bsky.social and I have talked about the pros/cons of Claude Code chats' default to ephemeral. We could consider baking this option directly into the skill

5 0 1 0
5 days ago

I'm still holding my breath watching, and man it's good to see Jayson Tatum playing basketball again

1 0 0 0
5 days ago

And there's more!!

@mcmullarkey.bsky.social and I got together to record a special Developer Science Office Hour where we talk about how our experiences as both psychologists & technologists informs how we think about dev tools, learning in the era of AI, & our collab πŸ₯‚

youtu.be/eW3OwDtmaDs?...

17 1 1 0
5 days ago
a shiny app with code on the right in the Positron IDE. Shiny app has an action bar on the left with "Upload" and "Analysis" with a file input and geometry input and a layer summary. The middle is an interactive map with puple and orange points. interactive sidebar dashboard visualizing penguins dataset.

{calcite} v1.0.0 is out πŸŽ‰

Oh and {mapgl} has esri-basemap support now πŸ‘€

Try it out with
calcite::open_example("page-sidebar-penguins")

πŸ€– using Claude? Feed it calcite context! r.esri.com/calcite/llms...

A release post coming soon....
#rstats

21 6 1 0
5 days ago

So glad that validation is happening!

2 0 0 0
5 days ago

Extremely relatable!

2 0 0 0
5 days ago

Wait this sounds so cool! Definitely excited to try when it's available

2 0 1 0
5 days ago
Linear walkthroughs - Agentic Engineering Patterns Linear walkthroughs - Agentic Engineering Patterns

Bonus, if you read the @simonwillison.net post on Linear Walkthroughs, thought to yourself "I'll get around to doing that", and haven't yet?

Orient can do that with "showboat" mode! Call /orient showboat in a Claude Code session after installing the plug-in

4 1 0 1
5 days ago

Had an absolute blast putting together this "orient yourself to a repo" skill with Cat, check it out and let us know what you think!

Great way to take advantage of the learning-opportunities skill on new repos (And let's be real, old repos where I've forgotten what on Earth I was thinking)

12 2 1 1
1 week ago
Preview
Kansas revokes driver's licenses of hundreds of trans people, prompted by new law Prompted by a new law, Kansas began sending letters to some trans people saying their drivers licenses were being invalidated until they change their gender information.

🚨hey so this is absolutely HORRIBLE for so many reasons I can’t even fully list out. But a few:

- licenses are immediately invalid, making it difficult to even get around to change license or god forbid APPEAL THIS DECISION

34 7 3 2
1 week ago
Preview
Do You Agree? Do You Strongly Agree? The Effect of the Number of Response Categories on Response Processes and Verification of Substantive Hypotheses Abstract. This study investigates how the number and labeling of response categories in survey scales affect respondent behavior, psychometric properties,

This is consistent with earlier psychometric work that suggests 5-7 is the best response scale options, but good to see that the finding holds up in contemporary research. Also, good to see that labeling scales whether anchored or not has little impact on findings. academic.oup.com/ijpor/articl...

116 40 1 2
1 week ago

In my experience I'm much better off starting a new session + re-providing context myself if I reach compaction

1 0 1 0
1 week ago

Had a good convo recently about this exact phenomenon! Writing in the "teaching" style can force us to confront what we don't know.

2 0 1 0
1 week ago

I feel this lesson in my bones. I'm someone prone to "thinking out loud" in writing. Micro-blogging helps me refine those thoughts

7 0 1 0
2 weeks ago

I whipped up another learning related Skill! Smaller than Learning Opportunities, but very complementary: interactive guidance through a quick research-backed psychological intervention that helps improve learning plans, motivation and commitment



github.com/DrCatHicks/l...

57 19 2 2
2 weeks ago

Congratulations!

2 0 0 0
2 weeks ago
A basic chat interface on mobile with messages about to do items, notes, and searching those notes. Most of the chats are slash commands

Might be up to something

0 0 0 0
3 weeks ago

This is my "desert island," most distilled, most succinct piece of advice right now for all the technical people I am talking to who are worried about their learning.

30 9 0 0
3 weeks ago
Preview
Data Consultants

I have continued to update my Data Consultants table. It currently includes 96 businesses. So if you are looking for help with data wrangling, visualization, analysis, reporting, dashboarding, training, and more, consider checking out these consultants!

docs.google.com/spreadsheets...

29 8 2 0
3 weeks ago

I'm sure summiting Mount Everest feels great, but have you ever had a delicate series of stacked PRs all merge smoothly?

3 0 0 0
1 month ago

Just made computer do hard thing 5x faster, I will never tire of this feeling

8 0 1 0
1 month ago

I have plenty of reservations about the reliability of funnel-plots and z-curves and such, as well as their interpretation.....

But holy shit look at that.

45 16 7 1
1 month ago
A photo of Weakened Friends on a lit stage in an otherwise dark concert venue

If Weakened Friends plays a show in your city, do yourself a favor and go see them

1 1 0 0
1 month ago

Somewhere inside me is a YouTube video essay about data grifting

I'm not sure if that's great, terrible, or both

4 0 0 0
1 month ago

That's not a small investment! And it might not pay off! But someone is going to try something like this approach and have a bonkers level of success

0 0 0 0
1 month ago

Anyway, you can't answer OP's super important questions without better measurement, there is no free lunch

Think creating + validating custom evals for your codebase. Then tracking if improvements in those evals track at all with fewer 3 AM pages, more on-time delivery, etc

1 0 1 0