HN user

framebit

576 karma
Posts3
Comments142
View on HN

I found the two Sam Quinones books on the US opiate epidemic fascinating.

- Dreamland: The True Tale of America's Opiate Epidemic

- The Least of Us: True Tales of America and Hope in the Time of Fentanyl and Meth

I'm an ML Engineer who's really closer to an MLOps role. I'm weak on ML and strong on data, scaling, cloud stuff, infra as code, making processes not suck, kinda everything _but_ the ML. So take my opinions for what they are worth, and keep in mind that the role of ML Engineering at company A != ML Engineering at company B.

I've described ML Engineering as putting the "science" in Data Science because we help introduce reproducibility. For example, I can take your model training and make it a robust process that happens over a huge amount of data on a daily basis with all the monitoring, logging, and reliability stuff surrounding that.

Some topics I would personally want to see for an ML Engineer on my team (and again, "ML Engineer" has less of a solid definition across the industry than "frontend engineer" or other roles that have been around longer) - Docker: can you containerize your code? Can you interact with a local container? - Model serving: at a basic level, can you wrap an API around a model? There's lots more systems design stuff here if you want to go deeper on model serving platforms. - CI/CD: do you know what Jenkins does? (Or equivalent) Can you talk about a coherent code testing strategy for ML code? How would you deploy a model service using a system like Jenkins? - Cloud stuff: you don't need to be an expert, but can you interact with cloud APIs directly or through Terraform, spin up instances, know the difference between object storage and databases, and do you have some Kubernetes experience (run a pod, get the logs, take some debugging steps when something's wrong). - Modern MLOps: model registry systems like MLFlow, feature stores (DIY preferred but vendors ok) - Scheduling and Pipelining: Airflow, Vertex Pipelines, lots of options here but those are the biggies. Know how to use these for basic data pipelines, model training, service deployment, and why and how you can deploy these via CD - Monitoring: know the difference and have strategies around monitoring systems metrics (cpu usage, etc) and model metrics (data drift, etc)

A lot of this stuff is harder to learn on your own because it often comes up in the context of larger teams and enterprise scale, where monitoring and reliability turn into KPIs that execs look at, but this is, to me, the stuff that defines the difference between a Data Scientists and an ML Engineer.

In some ways I feel lucky that tech was an early career change for me. I was trying to get a career in the arts off the ground, and it was going poorly. I was a recent grad millennial in the economy of the 2008 recession.

The decision to turn my back on what I thought was my passion was a profound spiritual experience. The decision to change came from outside of me. The decision of what path to follow was up to me though.

Tech was hiring and hiring like crazy, and I wasn't going to do an unprofitable degree twice so CS it was. I had a job before I graduated making 4x what my mom was making at her non-profit admin job.

If I hadn't pursued my art career first and had the chance to get deeply disillusioned with it, I would definitely be sitting at my desk trying to write code and thinking "what if... I'm not made for this... there's something else..." The truth is that I'm not cut out for the arts industry. I like stability, I like being salaried, I like having the upper hand in the hiring market (I know Big Tech is doing layoffs, but try spamming applications for a year to everything you can think of until the only place that calls you back is a cashier position at a grocery store. I have skills that are in demand now.) I like work that is decent and stimulating enough but which is definitely not "my passion" because that helps me keep boundaries on it.

I feel for folks who didn't get that chance to try out that other thing, who went straight into this career maybe because they wanted to, maybe because they didn't have the safety net I had that allowed me to do a second degree, maybe because life has held them down and change doesn't feel like an option. I've been out there with my chosen field and gotten burned hard by it so I'm content to stay put. It's definitely one of the cliche sayings about how the lows make the highs much higher.

I have no useful advice for anybody beyond their very early 20s facing this question. I know I would be eaten by this question if I hadn't already gotten my answer at the start.

The connection of lead poisoning and narcissism does seem plausible.

It's also worth noting that, broadly speaking, the parents of Boomers (depending on their age) went through 1 or 2 world wars, the great depression, the spanish flu, etc. and probably didn't come out unscathed. The trauma of the parents is bound to have ripple effects on the mental health of the kids.

There's more than a few stories of dads who went off to war and came back silent, drinking, and traumatized.

I avoided learning anything having to do with front-end for a long time, and I'm just now catching up. I've been doing a course on Responsive Web Design with scrimba.com, and that starts with a really comprehensive overview of CSS basics, including margin collapse, which I'd certainly been frustrated by in the past without having a name for it.

Just a little local perspective on the lines in Georgia. The waits that some folks had to endure in the primaries were a huge problem. Georgia absolutely has history of voter suppression like many post-confederate states, and some seriously shady stuff has gone down in the past.

However, the really long lines yesterday were more attributable to incredible turnout on the first day of early voting, which also happened to be a federal holiday so many people had the spare time to come out and vote. There are new voting systems, new volunteers for whom that was their first day running the polls, and a very motivated electorate.

For comparison, today, two days later, I know of at least two people who were in and out in 15 minutes in some of the same areas that had incredible lines on Monday.

There are legit voter suppression stories of great concern in GA, but I don't think the lines of Oct 12 qualify. I'm totally open to being corrected on this opinion though.

Oh wow, I had no idea Scrimba was brand new!

I signed up for the Responsive Web Design course last week and I'm getting a lot out of it. This is actually my first time taking any kind of web course of any type, programming or otherwise.

I work in the data/ml stack, not the web stack, and I've avoided learning anything serious about frontend for _years_. Figured it was about time to fix that so that I can get my personal website to look a little less terrible.

The Scrimba course looked like it would check all my boxes about technical content without forcing me to a timeline or a certificate program or being crazy expensive. I'm not looking for a career change into frontend, I'm just somebody who thrives in a structured learning environment for a new subject area.

I can definitely recommend the lesson + interactive in-browser editor combination. It's a powerful tool for fast, real learning. And I'm very grateful that somebody finally explained CSS margin collapsing to me!

My biggest critique so far is that the preview window for the project doesn't really work in Firefox, but things seem fine in Chrome.

How do you define success for you life? Put differently, what do you define as The Good Life?

You have to define that metric for yourself. Nobody else can do that for you. That said, I can offer a personal illustration.

I was in an art-oriented career previously. I defined The Good Life back then as achieving prestige in my field of art. Several years in, I came to realize that, for myself, my definition of The Good Life had changed. It now included time to spend hanging out with and helping friends; getting to know my nieces and nephews; having time to pursue art that's meaningful to me without attaching my paycheck to it.

I've been able to make progress towards my personal definition of The Good Life by changing careers into tech where job security is more guaranteed than in the freelance art world, and where the jobs are interesting enough and pretty cushy. That's given me the financial freedom to have that balance in my life with time for friends, family, religious life, art, my spouse, and other things.

What is success to you?

FWIW, I'm not "buying" ebooks or audiobooks anymore because I found out about the magic of digital access via a library card. This is certainly not a new thing, but since it'd been years since I engaged with the public library it was new to me!

I've been tearing through material lately, all through the library app (Libby, which I think is a rebrand of Overdrive), and it's still magic to me that I'm getting all of it for free. Sure, it comes with the normal library restrictions (popular stuff has long hold waits, sometimes tough to find a specific book vs. any book, etc.) but the sheer amount of books available that spark my interest is really cool.

I fell out of the habit of reading actual books for awhile, and diving back in has been like reconnecting with a long lost friend.

There was an article in the NY Times a few days ago from a respiratory doctor who explained that the nature of the physical mechanisms in covid-19 pneumonia mean that the patients' oxygen saturation will go way down before they start feeling short of breath, which accounts for the folks showing up at the ER with ridiculously low saturation. He noted that the celebrities who have recovered from covid-19 generally have really close, constant monitoring of their oxygen levels so problems are detected and treated way before ventilators are needed.

[1] https://www.nytimes.com/2020/04/20/opinion/coronavirus-testi...

My public magnet high school was a weird setup where the kids had a huge variety of educational backgrounds.

It was striking to me to be alongside kids who were naturally intelligent and gifted but who had gone to really crappy elementary and middle schools. I remember a math class in 10th grade where the boy next to me was really struggling through his work and shyly asked if I knew what he was doing wrong. I, in my arrogant 10th grade way, told him "don't you know about the commutative property??" And he gave me a blank look in response. When I explained it, he immediately understood and flew through the rest of his work with no issue. That was when it really hit me that I was no smarter than him, I just had the privilege of an elementary and middle school education that didn't suck. That boy has gone on to get a Masters from Yale and I think he might be working on his doctorate now, but he had to work hard to fill in some fundamental gaps.

I'm truthfully alarmed by the "free college" debate because of credential inflation. Our high school diplomas are nearly worthless, a bachelor's degree has become compulsory for a lot of careers that shouldn't need it, and graduate school is leaning the same way. I would rather see us as a country fix elementary and middle school education so we can make high school education meaningful again. While we're at it, my county took away stuff like auto shop and any kind of trade school class, which is ridiculous and even criminal IMO.

Of course, so much of what we talk about when we talk about education is inequality. In the haunting (paraphrased) words of one teacher that I heard on a podcast, "How am I supposed to teach them long division when they haven't eaten in days and they're going to get raped tonight when they go home." That whole "breakfast is the most important meal of the day" study didn't control for household income, so all it really showed was that kids who are food insecure do way better on standardized tests when they eat anything at all beforehand.

There was also an interesting article in The Atlantic[1] recently discussing the really strange US practice of teaching skills instead of knowledge. The idea is that kids learn how to learn, but the outcome is that they're horrifically confused and don't know what or why they're learning anything. When teachers switched back to knowledge based curricula, the measurement of skills went way up.

I suppose this is a rambly non-answer. I grieve the state of public education in the US where so many good people face so many barriers (poverty, clueless administration, their own food and housing insecurity, etc.) to doing good work.

[1]https://www.theatlantic.com/magazine/archive/2019/08/the-rad...

As a bike commuter who's not exactly in Tour de France riding condition, I whole-heartedly love what few physically protected bike lanes there are in my city. They've greatly extended the range of places I feel comfortable riding my bike. I don't know if I'll ever be in good enough shape to feel comfortable claiming a lane and riding in traffic anywhere but neighborhoods and steep downhills. Yeah, it's a pain to turn left from a physically protected lane on the right, but those small delays are totally worth the added safety of the curb/whatever.

Exactly right, it's a human nature twist on misinterpreted Protestant theology. Or rather a human instinct dressed up with the language of religion, in this case Protestantism but it could be anything. The manipulative language of MLMs which make money off the failures of their consultants, pulls from this work = salvation misinterpretation that boils down to "you get what you deserve," which is completely at odds with _real_ Protestant (and Catholic, orthodox, other Christian) theology.

IMO the Protestant work ethic, while it has its roots in the ennobling of work in service to God, quickly got twisted through the Puritans and their adherence to "working out your salvation" rather than "for it is by grace you have been saved, through faith, not by works so that no one can boast." Even if that wasn't the official theology, that's certainly the behavioral tradition. And this quickly got twisted into the fundamental American myth: you get what you deserve.

Are you wealthy? Then you must have worked hard. Are you poor? Then you must be lazy. Did participating in an essential oils MLM bankrupt your family? You just didn't want success enough.

So the protestant work ethic in a modern sense, (IMO, of course) is ennobling work BECAUSE your "righteousness" for whatever religious or non-religious definition you want to use is based in your hardworkingness.

A therapist friend recently attended a conference where a big group of doctors and health professionals focused around adolescents presented on the link between pot and the development of psychosis and even schizophrenia in users under age 32 or something like that.

I've definitely heard about this colloquially with people reacting very differently to the same strain ("we were all fine but she was just freaking out"), and the new science coming out of places like Colorado seems to be showing a strong correlation in that direction. I guess you could say (colloquially again, since I'm not at all in the medical field) that pot can open a bad door in some people that may have stayed shut otherwise. Is that gene expression? Brain plasticity? Dunno.

I think prohibition does way more harm than good and I'm glad to see legalization coming to the states, but I don't think we can treat pot as totally harmless for people under 30 and teens in particular because it seems like it's very harmful for some.

One paper on the subject: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2931552/

As somebody who made this career change, do not underestimate how much she has to learn. It's very doable, but there is really SO MUCH. For somebody who didn't grow up tinkering with computers (I didn't, my Dad needed that desktop to write his law briefs so if we messed it up we were in big trouble) there's so much ground work to lay and jargon to conquer that is second nature to people who program but very confusing and new for beginners.

For example, early on when I was just starting to program, somebody was trying to help me use a terminal (a what?) and they said "ok, type L S dash L pipe grep dot txt." HUH? ok, I typed "ls-l," but what's pipe? No no no, it's "L S Space dash. ls is the command." It's a what? It's like software that's already been written that you can call just by typing the name." Oooohhh.

That's just one example. I got it eventually, I got a CS degree and I'm doing distributed systems stuff fulltime, but man those first few months were rough. I remember a friend helping me through what I later recognized as Baby's First Off-by-one error. I remember the victory of writing my first program that worked the first time. I remember the first time a classmate asked a question and suddenly I was able to answer it!

My recommendation to start: write a toy program to do something trivial and run it from the terminal. Like "Hello ___" where you take in ___ from the command line. Learn how to interpret or compile your code from the terminal, learn the basics of variable assignment (== vs =), if statements, maaaybe loops, and go from there.

Never, ever say "It's simple," "It's easy," or "It's trivial." I got very hung up (still do sometimes) on trying to understand what I'm doing from code down to bare metal, and that is NOT SIMPLE, and when you're swimming in unfamiliar terms and confusion it is very hard to articulate a question that clarifies your confusion. And trying to guide somebody into thinking on a handful of abstraction levels without going to far is not simple. Many times I needed to hear "you don't need to worry about that right now" or "just treat this as magic for the moment and focus on this other thing." That is way more helpful and encouraging than "it's simple." I think FizzBuzz is simple now, I sure as heck didn't back then.

Like other commenters said, leave OOP or functional paradigms or whatever for much later. That stuff matters when you realize your one class has gotten super unwieldy.

Don't jump too quickly to Android or web development without covering the basic basics first. Android and web development are not simple and there's some black-box magic that I don't even think some pros fully understand. It is nice to visually see the changes that you're making through editing a UI, but the abstractions can be tough.

This comment got long. In short, be encouraging, help her discern what she needs to know and what she doesn't, and nothing is ever simple because ultimately it's all running on a lightning-infused rock, and these days that rock is probably connected by metal strings or electro-magnetic waves to other lightning-infused rocks, and that's pretty hard to wrap your mind around.

All that said, I know for a fact that she can do it because I did!