morelandjs

Recent, I built a new tree-based ML algorithm, and the nature of the problem required that I write “non-pythonic” code.

I discovered that ChatGPT is actually quite good at converting Python to Cython. Strongly recommend if you haven’t tried it!

He’s ignoring the important details by using “exchange labor for compensation” in an outdated fashion. Money begets money.

How else does one explain the concentration of wealth over time? Progressive income tax is needed to stabilize wealth from collapsing in on itself.

I’ve used it as a programming assistant and as a general question-answer assistant and it’s so massively useful despite its imperfections I wonder if I’m using the same tool as people with comments such as yours.

Currently at Amazon, did a physics fellowship with the DOE and completed a practicum at LLNL. I decided against the labs mainly for location reasons. I have nothing but nice things to say about the DOE labs and the people that work there.

The pay is good for government work. You can continue to do fundamental science research. You get great benefits, and they do some wicked cool science and engineering there (read about NIF and Z-pinch). I didn’t get to see any of the wild stuff behind security, so I can only imagine.

The skill I value most in a DE is the ability to build effective model deployments and project layouts that are SIMPLE.

If I were to switch functions and find some protected time, I’d go off into the woods and build an example deployment compatible with company infra that is as light weight as possible. Then evangelize my team.

This is the truth. Model tinkering in Jupyter notebooks is heavily saturated and won’t get you very far now days.

If you know data engineering and a little devops as well as graduate level stats and linear algebra… the jobs are bountiful.

Dali Oberon 1, Audio engine N22, and Kanto SE6 speaker stand.

Edifier has nice budget options if Dali is too much.

Very true. Golden handcuffs is an option, but you can also just liberate yourself from financial stress and help the people around you who need it. Not everyone is chasing a BMW or bragging rights.

Specifically, the study, published in Nature on 21 September, shows that just 20% of PhD-granting institutions in the United States supplied 80% of tenure-track faculty members to institutions across the country between 2011 and 2020 (see ‘Hiring bias’).

Nice pareto principle example

For starters, they are not accurately advertising the compensation for the gig. By every sane measure, the money earned is wages less the cost of owning and using the vehicle. They are not so subtlety convincing people to sign on to a terrible deal, because a lot of workers don’t realize how expensive vehicles are.

Exactly, if you can falsify theories subject to the assumptions of the simulation, that’s still evidence.

I've been wondering for some time if it would be possible and effective to delegate all administrative and process responsibilities to line managers. Remove the IC from the planning documentation process entirely and let them plan their own execution however they see fit within each Sprint.

The only responsibility for the IC would be stand ups with their manager to discuss progress and agree to future timelines. The ICs wouldn't have to go to any other Agile meetings or write any business/project planning docs. Manager would scope and write all IC tickets.

I think you would very quickly see a culture change with regard to process work. Having sole responsibility for both creating and completing the process work, it would be in the manager's best interests to make it efficient.

I've never seen an org try the above, but it doesn't seem crazy to me. I expect you'd still be able to attract managers through increased compensation and opportunities for career progression.

Note: I'm not saying the IC forfeits all planning discretion, I simply mean it would be the manager's responsibility to write it up into whatever documentation is required by the org, track progress, rewrite the plan as necessary, etc.

China has a lot of brilliant scientists. They also have 1.4 billion people. Not surprising.

What are my options for big data that won’t fit completely into memory? Is it easy to hook up to a spark cluster?

Do I have the option to access the underlying infra through a Unix shell, when the UI isn’t enough?

I suspect you are greatly over estimating how much this has disadvantaged you in the grand scheme of things. Move on, forget about it, and count your blessings.

What happened to you is wrong, but it’s also a daily occurrence for many people that are not white men.

I don't think your estimates are accurate. A comfortable 2 bedroom can be had in the nice parts of Brooklyn for 3.5-6k. Manhattan is somewhat more, but not drastically so.

Grew up in the midwest and moved to Brooklyn from Philly. Absolutely love it here. City is truly alive with energy and the job prospects are fantastic. Your mileage will vary though depending on how much you make.

If you are in undergrad, I'd recommend something like,

Math or math-heavy science BS -> undergrad research -> computationally heavy PhD -> entry level DS or ML engineer job -> senior ML job (within a year or two)

If you are older and looking to pivot, I'd recommend, Data engineer -> senior data engineer -> entry level DS -> senior DS

Don’t agree with some of the statements of the author. Delivery times have gone up? I’ve had quite the opposite experience. I’m frequently getting same day shipping. And this is in the midst of unprecedented logistics challenges.

Advances in additive manufacturing. Think 3D printers on steroids.

Happiness is the distance between reality and expectations. In a healthy society, you raise your expectations over time. In the US, blue collar workers are not seeing gains and in many cases are losing ground. That’s why people don’t want to work.

I work in this field. Author makes great points. Here are some additional thoughts: * Don't use bad optimization to discredit all optimization * Incremental A/B testing assumes that all accumulated features are independent with no interactions, which is almost assuredly a poor assumption as a website grows and space becomes limited * A/B testing is the only reliable tool to test hypotheses; causal inference is great but not a substitute * Incremental A/B testing (small changes) should be periodically coupled with mutational A/B testing (large changes) to tunnel from local optima to global optima

Strongly agree with the author. A logistic regression built on top of perfectly crafted features will out perform more advanced approaches with crude data.

I never viewed full stack as knowing all the technologies. Rather, I view it as the ability to solve the problem end-to-end. Often, knowing a few generalizable recipes is sufficient to be effective in this regard.

I don’t believe this kind of statement in general. It’s a whitewashed way to say “I’ve tied up my self worth in money. I like keeping score and reminding myself of where I stand relative to other people “. In short, it’s ego.

But no one wants to say that so they make up stuff about “providing for their family”.

A common theme I see, is tech venture capital types without any physics background complaining about the over regulation of nuclear energy. If there's one technology you'd rather over regulate than risk a catastrophic externality, it's nuclear.

I'm just imagining what a shit show it would be if the "move fast and break shit" Silicon Valley mentality was applied to nuclear energy. No thanks.

I would strongly prefer a cautious but ambitious government plan spear headed by physicists from the national labs.

numpy started a science renaissance. fantastic library for what it does.

Question about this. The career path to high level decision maker seems to pass through JIRA manager. Why is that? Are there other parts to gain decision scope within a company that don’t require a stint in management first?

In my experience, you will not get anything productive out of such discussions. When your salary drifts too far, go audit your salary with an external offer. Ask for a match or leave. Spare your current employer the drama.

Just to add that not all ML positions are Deep Learning. Still good info to know though.

Exactly. A lot of amazing math has come out of string theory.