My dad ran a job shop focused on small jobs and the economics are different.
A lot of his work was keeping other local shops / industrial equipment up and running. So there is a lot of variety of work but very low throughput and kind of by deffintion you have the capabilities to fix your own machines.
Programing a CNC machine makes it east to make a lot of the same part but if you only need one it may be quicker to just knock it out manually.
A 50 year old mill or lathe is easy to keep up and running, can be upgraded with a Digital readout or even CNC controls if desired. A tool in a shop like this likely won't see the cycles one on a factory floor constant uses sees but may be worth keeping around since it offers a unique capability...he had a large ww ii surplus lathe for jobs that wouldn't fit on the smaller more modern machines for example.
I've eaten at Dave's and also gather and eat morel's regularly in Montana.
Dave's is a popular joint but didn't strike me as particularly good even by Sushi restaurants in Montana standards. They had some creative rolls but seemed to fail to get the basic fish cuts and rice seasoning right.
What they were doing marinating the morel's and eating them basically raw is ridiculous. No one who is experienced with morel's does it that way and most mushroom guides contain info on hydrazine contents/risk and suggested cooking methods.
March and April is too early for Montana (or most US) Morel's which are commercially picked in forest fires from the previous year once the soil temperature reaches 54 degrees. I heard at one point these were grown in China and wouldn't be shocked if they are a different [sub]species that might contain more hyrazine. There are also some "false morels" that fruit earlier (one species is known as the snow morel as it comes up right after snow melt) but these are easy to identify contain even more hydrazine.
The g7 is definitely a bit better on all counts except maybe bluetooth reliability. I use them with [open source closed loop](https://loopkit.github.io/loopdocs/) insulin pump software and it is pretty great.
Still a lot of plastic but the sensor comes in a smaller jar/applicator in a cardboard box and it pretty reliably lasts 10 days cutting down waste.
It sounds like this new sensor is just a g7 but with less stringent accuracy standards for non insulin users allowing it to last 15 days.
I actually find hydraulic brakes much easier to maintain. They self adjust to account for wear and don’t suffer from cable stretch so I rarely need to adjust them. They need a bleed maybe once every few years but that is actually dead simple for shimano anyways.
The outdoor industry as a whole had a covid boom and upped production/orders. Now it seems a lot of those sales were in a way borrowed from the future and they have a ton of inventory and lagging sales.
It isn't even that much of a speed hit using the classical sorting CART implementation. However xgboost and ligthgbm use histogram based approximate sorting which might be harder to adapt in a performant way. And certainly the code will be a lot messier.
Is anyone seriously proposing removing dams on the Colombia or Snake? Or really any dam used for irrigation?
Most of the removals seem to be on very small scale hydro projects on coastal river that produce very little power, aren’t used for irrigation and open up a lot of salmon habitat.
Gpus were built for dense math and they ran with it. To the point current best architectures are in part just the ones that run best using the subset of linear algebra gpus are really good at.
There has been a lot of work on sparsity and discovering sparse subnetworks in trained dense networks. And intel even proposed some alternative cpu friendly architectures and torch/tf and gpus are starting to do okay with sparse matrixes so thing are changing.
My step mom did this in the 70s or 80s. She said it was the first time she heard primal screams and that you could just pick up high end outdoor gear along the trail as people would discard it to save weight.
There are tons of successful long term non religious cooperative living arrangements of various sorts but yeah most not fully isolated substance farming because that is hard.
Amish communities also aren't self sufficient though. They'll happily sell and buy goods with outsiders and even sell amish furniture via websites (not sure if they use an intermediate). A ton shop at the Costco near me.
Propane is a bit more expensive then traditional electric heat depending on the year.
It used to be much cheaper so a lot of old houses that aren't near a natural gas line were built with propane heat. You might save $200-1000 a year by switching to a heatpump but that means 10+ years to offset a ~$10000 install cost for a large unit professionally installed with a new 220 circuit run for it etc so most people just keep paying for propane or maybe use small electric space heaters.
And if you are somewhere you need to worry about pipes freezing in power outage a while you are away or temps bellow the heatpumps minimum -15f range a propane stove is a nice option. Relatively cheap to install and they can be setup to run on a thermostat with no grid power.
If i was building a new house i'd go heatpump + a propane stove for back up heat and a dual fuel induction/propane range.
It’s not bad where I am now in rural Montana but where I grew up near Seattle the power was off for a week plus most winters.
Lots of above ground power lines in heavily forested suburban sprawl means lots of branches in lines every wind storm and they (private) electric co has to visually check all of them before they can turn power back on.
Despite being much more rural the power coop that serves me now seems to never have an outage that lasts more then a few hours. Probally mostly because they have much less pine to check.
It needs good air flow to work well and the noise is only really a problem inside the house in the room next to it. I may put some sound insulation on the wall at some point.
We are running a large single head Mitsubishi ductless h2i in Montana. It easily provides most of our heating needs down to about -5F and some heat down to -15F, we have our old propane boiler / in floor system that kicks on when its really cold.
And it provided ac in the summer which we didn't have previously.
There are some downsides, it was expensive to put in and the outdoor unit is loud. While it does save us money, it probably wouldn't make sense as something that will pay for itself in heating costs with current propane prices and natural gas is much cheaper if you can get it. But definitely makes sense for new construction especially when paired with a wood or propane stove for back up heat in power outages/ambiance/quick heating.
Not super common but its reasonably frequent for someone to go to industry for 2-3 years and then go for a masters/phd.
Those candidates tend to be great as they are up on industry practices/technology like source control and databases where as some phds can have only academic coding experience. And they tend to have studied something they really knew they were interested in and really been the driver on their thesis project vs just contributing to their advisors research.
You also see great phds who didn't wait but really took ownership of their project, used source control, figured out distributed computing, contributed to open source, scraped/built their own datasets, understood the real world implications and hacked on side projects to develop coding skills.
And you see some who just completed a theoretical + computational project their advisor suggested on an existing dataset with the minimal amount of coding needed and little thought to implications/applications.
I've been in the ML/data space for 20 years and went through the 2008 cycle early in my career. I'm not hiring at the moment but am always happy to review resumes and give career advice as part of networking so feel free to ping me.
Reaching out to managers at big cos won't get you much. They are getting a ton of twitter/etc candidates and have stricter hiring policies and rubrics to prevent nepotism and favoritism. If you can network and ask for referrals through your friends / professors that can work better. Or reaching out to early stage startup cofounders can work really well.
ML/AI is less frozen then some other areas so that is good. Most really competitive new PHDs will have either a couple of internships, strong academic contributions or previous engineering experience demonstrating strong coding ability.
You should also consider post docs or academic engineer postings. The grant cycle insulates these a bit from the economic cycle and they can be a good place to gather some experience while you ride out the cycle.
And definitely consider very early stage start ups. A startup that just raised and has 2 years of runway is probably one of the safest places to be at the moment as they are still focused on growth. A lot of great companies proved themselves as startups during the 2008 cycle and grew rapidly after. Networking can mean a lot more here as early stage founders often literally just hire their friends or people they get along with without a ton of process.
My dad ran a job shop focused on small jobs and the economics are different.
A lot of his work was keeping other local shops / industrial equipment up and running. So there is a lot of variety of work but very low throughput and kind of by deffintion you have the capabilities to fix your own machines.
Programing a CNC machine makes it east to make a lot of the same part but if you only need one it may be quicker to just knock it out manually.
A 50 year old mill or lathe is easy to keep up and running, can be upgraded with a Digital readout or even CNC controls if desired. A tool in a shop like this likely won't see the cycles one on a factory floor constant uses sees but may be worth keeping around since it offers a unique capability...he had a large ww ii surplus lathe for jobs that wouldn't fit on the smaller more modern machines for example.