From the historical point of view, this heavy decoupling is associated with microkernels which fell out of favor with the rise of the Linux kernel. I think in this day and age, modern programming languages could make building a microkernel that doesn't fall prey to the shortcomings of MINIX a possibility.
Microkernels are IPC-heavy, right? What I rather imagine is having modules at source level, not kernel level. So the thing still compiles down to a monolith/hybrid, but modules are abstracted away and reusable.
The IPC overhead is very much manageable, microkernels tend to be a lot more responsive than monoliths and with paging for message passing the overhead is reduced even further.
Microkernels are much easier to optimize for multi-core CPUs than monoliths. The ´kernel modules´ from a monolith run as user processes in a microkernel environment so they automatically benefit from more cores.
I would say, with your background, consider applying to DeepMind[1], or a national research lab. What you'd be looking for would be "entry level" research work. This would allow you to easily stand out in your applicant pool, and make it to the interview, where'd you'd have a better chance to properly explain your non-traditional academic background. If you're a US Citizen, maybe consider applying to positions at the MIT Lincoln Lab, Los Alamos, or Argonne National Lab. A lot of these guys aren't doing that much with ML/RL/AI, but that's okay! If you're serious about research you'll probably want to do a PhD. To get into a good PhD program, you'll need good references.
P.S. I specifically underline references and research experience because that's where it appears you'd be weakest at the moment. Getting a 4.0 GPA in your grad school coursework (which is online, I presume) says a lot more about your organizational skills (which must be really good!) than your ability to research. Getting someone with a PhD to write "This person is capable of doing good research and I've seen it with my own two eyes" is probably one of the best things you can get on a recommendation letter to a PhD program, which is far more valuable than "this person did well in my class" (see: https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&c...)
The audio quality of these interviews made me grimace. Audio quality aside, this looks like it would be a great resource for anyone trying to go through a "FAANG-style" multi-round interview process.
Small nitpick: in most cases, these companies make a hiring decision with just one round of interviews (if you exclude the technical phone screen, which is a way to reduce wasting everyone's time when it's obvious it's a poor fit).
As both a candidate and as a hiring manager, I much prefer to have a single round of on-site interviews, vs. having the candidate come on site more than once. Good candidates probably have a good job already, and asking them to take multiple half days off to interview is an unnecessary burden.
Posts like these really continue drive home the message of "Silicon Valley is just a shell of what it used to be. Only go there for the sake of relocation or with a job offer in hand." There's nothing left for the dreamers or hopefuls in SV who have nothing to run on but just that--hopes and dreams. It just makes so much more sense to take a pay cut and swing for the fences in Atlanta, Houston, Austin, or Phoenix. Hell, I'll still throw in Boston and Seattle because they're still THAT much cheaper than anywhere in SV.
Austin is a great place to do this type of homerun attempt without experience. There are some nice start up jobs there, and the costs are way lower. I went from Austin - > sf and while I'm still in the career-building phase here in sf, it's not so desperate that I am living in a closet. So I was lucky that I came from Austin first with some experience already. But this is a really bad city to build something out of nothing. The costs are stupid, even when you see a rent like "only $1,200 a month!" it's likely that the deposit is ridiculous too. The transportation is great, but it depends on where you are in the city. I luckily get on the bus at the start of its route so I always get a seat, but if I was a few blocks down when the bus was already packed like sardines, I'd hate it.
Train is the same. You have to go out so far to be comfortable that your commute goes into the hours mark. SV is just insane for someone that doesn't already have an established career. I know I would have burned out already if I didn't have the experience from Austin already.
Phoenix is a good option if you don't mind the summers. The salaries are very good for the cost of living (salaries seem to be increased because of the proximity to California, I even somewhat frequently see startup jobs in SV with lower salaries than an equivalent position in PHX with some negotiation). There is a good mix of startups, mid-sized companies, and juggernauts like Amazon, Amex, PayPal, GM, etc.
Importantly in my opinion, you can buy a nice house on middle class income in Phoenix still, or rent for very cheap, and the traffic is pretty good. You'll be car dependent in most cases but it is RADICALLY cheaper than Silicon Valley.
Boston isn't really cheaper than The Bay. Traffic is terrible, so commuting by car long distances is out, and living near your job is often just as expensive as The Bay.
Can't really agree with that. There's still quite a few sanely priced neighborhoods with good transit access, and without huge crime issues. (Ex: Malden)
(True also for Los Angeles, though at least that has the Pacific Ocean)
This is the (not so?) subtle quality-of-life aspect that people seem to forget when mentioning other cities as alternatives.
The Bay Area isn't popular/expensive solely due to network effects. Climate and geography (nearness to ocean, mountains) make it comparitively more appealing than the alternatives, for a variety of ages and lifestyles.
Austin is way nicer than Houston or Phoenix, not sure about Atlanta. Acutally Austin/San Antonio pretty much has the nicest weather in texas. I'm from fort worth and fort worth gets both hotter and colder than austin - climate wise it's the california of texas (austin also has comfortable levels of humidity unlike houston).
This really drives home a sentiment I've acquired during the course of my college education: CS/Math is often considered "hard", but I feel that's just because we've struggled with getting good visual/verbal communicators to dedicate their lives to CS/Math education. I really feel that when explained properly (and the definition of "properly" sometimes need to be adapted from person to person), topics like Gibbs Sampling or Fourier Transforms or Backpropagation aren't topics that should take entire weeks of self-study to grasp in 2018. Yes, they require some math background, but there's some strong intuition behind them. Maybe I'm just slow or thick in the skull.
This is some really cool information! I implemented a map-parsing algorithm that used a data structure that enabled neighboring cell access in constant time. It's biggest weakness is that very high-resolution occupancy maps take quite some time to render. I'm very inexperienced when it comes to sharing my coding projects with the world, so any feedback on the readability/accessibility of my repository would be greatly appreciated! https://github.com/dwrodri/LQTLD3
I have not used real robotic simulations, but I have used RViz and ROS for testing my own path-finding algorithms. Performance on my Late 2015 MacBook Pro was stuttery (about 15-20 fps), but still good enough to be somewhat usable.
Overall, I had more issues with finding good resources on the rospy module than anything else. It seems almost all robotics development with ROS is done using C++ instead of Python (in my little experience), so there is little I could find in terms of code samples that would allow to get past a project like this (https://github.com/gandalf15/CS3027--CS5059--Robotics-Univer...).
I felt the same way during my lectures. I wanted to use Python because I had already implemented the fuzzy functions I needed for obstacle avoidance and it is overall simpler, but most people on class used C++ anyway.
FWIW, after realizing I never directly answered your question, I figured I should give you a full response.
Rospy aside, writing and implementing your own SLAM/path-planning tools in ROS is quite simple once you get familiar with the architecture of ROS. All the sensor data can be collected by subscribing to the right channel, and then you can return driving instructions back to a controller by publishing them on another channel. Since all this channel business operates on the network layer, you can even offload the computational workload of the algorithms to a remote system and have the robot get its moving instructions over Wi-Fi.
I think the best takeaways from the SO survey come from year-over-year analyses. You can watch JS explode in popularity over time, and see how quickly languages and libraries get adopted.
It would also be really nice if the "popularity" metrics that OP mentions (i.e. preferred language, least preferred language, language most used at work, etc.) could easily be grouped by "Developer Type". After some quick googling, it has become apparent that the data from previous years is available to download, but it would be nice for SO to show that data in the blog post.
I think this is the first game I've seen posted to HN that wasn't a "Show HN" post. That being said, I haven't been a member of HN for that long. I'd be very interested to see what else has been posted here.
