Welding is coming along. Ever heard of Path Robotics? That's high-mix, high-complexity welds. (There are a lot of other, less-sophisticated welding robots out there.) The biggest moat for a welder right now is special certifications to be able to do welds on submarines, etc.
I tried to build this back around 2020. I think my concept was trying to be too cute in several places. But this was the basic idea.
I found that neither side of the market wanted to rethink the market, they just wanted something that worked well for them. Even today, job seekers may be reaching for something like this but employers have no interest; there is a glut in the market.
I’m sure I never quite got the messaging correct. However, I distinctly recall that triplebyte attempted a pivot in this same vein and also failed bad.
I've been mostly pleased with our use of python-polylith [1] with poetry in a production application. We output a webapp, python sdk, and CLI as separate "projects."
It doesn't _really_ solve python dependency/import issues. Instead, it helps to you keep your project in discrete chunks and well-organized. It also makes it easy to package up the separate projects as artifacts.
I've run into some issues with versioning separate projects but I suspect that is a matter of bandwidth rather than an actual, insoluble issue.
I'd use it again at a startup or on project where you need to ship a bunch of artifacts but don't have a lot of bandwidth.
Are there any comparisons of weather apps by area? For example, "for the San Francisco bay area, apple weather is most accurate on rain. But for NYC accuweather is better." I suppose you ought to be comparing weather APIs rather than apps but it would be most usable if you just knew which app to download.
I would be careful with LaTeX. I use to have a LaTeX resume generated with LuaTeX. At an old company, I saw my LaTeX resume in the ATS long after I was hired. Apparently, something happened and the PDF displayed as blurred-but-not-unreadable in the ATS. Maybe the ATS did some post-processing or used a limited PDF display engine? Lucky for me, the resume for that job was just a formality. These days, I just use Google Docs and export to PDF.
I would have liked to see them focus on less well-known companies. It seems quite possible that the reference human responses (from exponent) were already present in the training data.
I would be interested to hear more about how much they were discussing their failures/challenges with others. It sounds like this might be a case of, "I finally talked to someone about it and they unblocked me in an hour."
When I was in grad school I was very hesitant to ask others for help or feedback. Big mistake! I see similar things with interns: they'll wait until a daily or weekly check-in to raise problems.
My approach now is to set a time budget; if I don't figure it out myself within X hours, then I have to ask someone for help.
I would love to learn if TripleByte's data was sold off somewhere? I also had this concern (which was probably a bit overblown). One of the main scenarios I was concerned about was a case where the firm was sold off for parts.
You could check out ethical ads [1]. I have no first-hand experience with them as a publisher. But they are fairly unobtrusive on readthedocs [2] sites.
Waymo is currently under investigation for multiple incidents, not all of which it had previously disclosed to the NHTSA [0]. The recent light pole incident also doesn't help [1].
If they are doing 50k rides a day, then they would appear to have a remarkable safety record.
It will be interesting to see if these investigations lead to a repeat of the Cruise debacle or if this will become the price of doing business.
Yeah, I wonder if that is 425k _formatted_ lines of APL? I'm blanking: wasn't there someone who wrote their hedge fund's entire data processing pipeline in 4 lines of APL?