dmillar

Replit is used a lot in this context. Their agent is good, but their circumvent-policies-to-get-something-in-front-of-execs-quickly is an amazing and mis-priced feature.

Low barrier services don't care who's first in this epoch.

You know it's big news when Gruber makes the HN homepage.

https://plaintextsports.com

- Enable RLS

and/or

- Turn off the REST API (if you just use pg connections)

- Disable the JWT/anon token(s)

Well, I see we have very similar hobbies and careers! Nice setup.

Strats as a role has been downplayed and realigned a lot over the last several years. Traditionally, a strat was indeed a "quantitative strategist" that sat on the desk, often next to voice traders, developing tools, pricing models, trading systems, etc.

One of the people I regard as the Godfather/OG strat is Emanuel Derman (https://emanuelderman.com/). Another transformative (in my time) person in the Strat complex was Marty Chavez (https://www.rmartinchavez.com/). Both amazing people on and off the trading floor.

Goldman has done this for decades, pushing it even further by having developed their own language (Slang), graph db (SecDB), and IDE (SecView). Many engineers resist working it in, but for any strat it's mandatory.

If this and/or the policies, leadership, and mindset that enable these kinds of things interest you, I highly recommend Abundance by Ezra Klein and Derek Thompson. IIRC, they outline data centers in space in his utopian introduction.

What the "right answer" is will vary widely on application and analyst. That's one reason there are so many coordinate ref systems. If you keep everything in 3857, you'll get answers in ~meters, but whether that's "right" depends on where and how large the distance or geom is and what your precision threshold is. So, really, everyone's needs are necessarily "specific."

DuckDB > geopandas, certainly for anything out of core. Though, I recently gave up on importing 70GB worth of large multipolygons (from a csv in hex wkb), and just used a postgis container. In concert with DuckDB's growth, I'd also mark the advent of geoparquet.

The big change, in my view, over the past decade in GIS software, is in compute and storage efficiency across the typical stack. DuckDB has become a part of this, but h/t to the advances from shapely, geopandas, geoparquet, and GDAL. There's a lot of overlap in that venn diagram, and credit should be spread around. QGIS is great, too, though I feel there is market opportunity to apply 90/10 to its massive feature set and move it to the web.

Many criminal records, petty or otherwise, are public record. When archived, expunged or dismissed infractions never truly become that. A traffic violation or other petty misdemeanor from 20 years ago, that has been expunged from official record, can show up on a background check because companies archive public data. So, there is a flip side to this.

Arturo AI | MLEng and MLOps Roles | 130-190k DOE | https://www.arturo.ai/careers

Computer vision (segmentation) and aerial imagery experience a big plus for the eng role. Remote/hybrid. Proximity to Utah preferred.

python, pytorch, sql, stats, linux, mlflow, aws

if interested, send me an email : dmillar at the above domain.

The UK still has a heavily embedded caste or caste-like social circumstance. Pharma and finance have somewhat eschewed this tradition, but even those industries still have its embedded lords and ladies. Many of their smartest head to the US (or elsewhere), and Brexit has set them back decades from achieving the "richest country" status. A software engineer can double their earnings by moving to the US (if they can manage to). The owners of capital are generally not creators, and until that changes nor will their economic station.

K&R

Arturo.ai | MLOps/Data Engineer | Remote/Hybrid | Utah preferred

- Python, solid ML lifecycle experience, computer vision preference

- AWS SageMaker

- SQL, data strategy, data warehousing

- Desire to work in highly dynamic/changing environment

- $150k-180k plus bonus/equity

Please read the JD on the site. If interested, please send your MLOps, SageMaker experience, and any other information you feel is relevant to the email below.

* Unfortunately we unable to sponsor H1B/transfers at this time

recruiting@arturo.ai | https://www.arturo.ai/careers

Would be interested to also know how they handle per-title audio. With stereo -> 5.1 -> 7.1 and the sides and wide layout variants, how do they think about this during the inspection and encoding process? Being completely naive to Netflix's source media, and assuming it comes in a variety for formats and media, it seems like there are decision to make there. Though audio obviously has a much lower bandwidth burden, one would think there could still be QoE gains (and bandwidth savings) by doing things that AV1 can do with different scenes in something like OPUS.

You're right. Thinking back, I think I asked "what's the average distance" not "what's most probable".

You have two random uniform variables and the distance/difference/change between to finite points.

Put another way (and code it up if you want). Select two random uniformly distributed points between 0 and 1. Do this 10_000 times, whats the average distance between the two?

This gets to the question of "most probable" vs "expected value". A conversation I always welcomed.

Early in my career, when I first started interviewing, I used to ask a version of this to recent grads. It was never a make-or-break question, but I found it to be a great way to a.) see how people approach problem solving and probability and b.) see how they respond when you start asking whys (even if they answered/guessed 1/3). It's something that takes zero code to answer, and the intuition is easy to grok once explained.

The other part I particularly enjoyed was the people who initially guessed wrong, but then got to the answer intuitively almost always sent me code proving the answer.

For the record, my question was: "Two points are randomly and uniformly selected on a line 0.0 to 1.0. What is the most probable distance between the two points?"

b/c it's slower

This is a bit old in SD terms, but I wanted to post here if anyone ever wanted to use a fine-tuned SD model to create a WSJ-style hedcut of themselves (or others I guess).

* I wanted to express that this work would not be possible without Noli Novak's incredible artwork, and attribution should be given to her when appropriate (and certainly not to me) *

Zola

It does what I need. Fast, search-capable, markdown, and handlebars. The community isn't huge, but it's just what I need to get a little bit of writing online.

"HIRWAY: ...Just music. It allowed some space for my brain to get bored. Another thing that I learned was how much boredom was, like, an essential part of creativity. Maybe boredom is too strong of a word for it, but it's something like..." mindfulness.

I remember hearing Sam Harris (I think) say something to the effect of 'once you reach some stage/awareness of mindfulness, you can never be truly bored.' Like, if you're bored, you're not paying attention. That's stuck with me.

Agree, but I do think there are real clinical benefits to short trials like this. For example, it's good to have evidence that one could quickly improve their cardiometabolic health immediately after a cardiovascular event just by switching to vegan.

The TLDR isn't a shocker:

In this randomized clinical trial of the cardiometabolic effects of omnivorous vs vegan diets in identical twins, the healthy vegan diet led to improved cardiometabolic outcomes compared with a healthy omnivorous diet.

Would like to see a couple of the middle ways peppered in (pescetarian, vegetarian).

As usual, Gruber was right on the money. Via Threads yesterday:

"My prediction is that Apple will make changes—fixing bugs and/or closing loopholes—that break Beeper Mini. It’s untenable that there’s unsanctioned client software for a messaging platform for which privacy and security are a primary feature.

It’s a very nice app, remarkably clever, and for now works like a charm, but if Apple wanted an iMessage client for Android they’d release an iMessage client for Android. Seems irresponsible for Beeper to charge a subscription for an unsupported service."

https://www.threads.net/@gruber/post/C0k1VgyMGZN?hl=en

Yes, it's short and insightful for all kinds of writers--technical and otherwise. The editions subsequent to the first have decreased emphasis on economics and instead geared it toward academic-style writing.

In-orbit refueling, etc. Those are Starships not SLS being counted: 29:45

Talking about methane, in situ refueling (not just spacex, but they've been the most vocal): 41:25

Complexity/redundancy: 46:28 (looong pause and laughter)

Comments on engine design: ~48:00 (you have to be aware of Raptor/SS booster design to get the criticism)

My TLDW: He's (not so) indirectly critical of Space X's Starship program and tribal groupthink. He may be "scared" of NASA because he's calling out one of their giant Artemis contracts.

Some longer takeaways:

- Simplify. We went to the moon 50+ years ago, but we are reinventing the wheel in some significant ways in Artemis. Why? We are supposedly going to the moon in two years, but we have never attempted a cryogenic refuel in orbit (this seems like a biggie).

- Communicate a lot. Why aren't people talking about the seemingly giant increase in complexity to accomplish the same mission we had in the 60s (land on the moon)? People need to have safe/comfortable way to raise questions and concerns.

- Have many layers of redundancy. Apollo had 6-7 backup procedures for what to do if they couldn't launch the lander off the moon.

- Test. small tests, big tests, real tests, skin-in-the-game tests. Some tests can be eliminated by simplifying or, in the case of big systems/tests, doing small tests on the riskiest components.

- Humans have ingrained biases. Will be situations astronauts haven't experienced where their instincts may be wrong.