schreiaj

Good news, in the US you don't need to choose... and by need, I mean you probably get both.

You almost certainly do not want an LLM to do that. Leap71 actually has computational models generating functional rocket engines that way. You could absolutely wrap a tool like that in a shell and handle control with an LLM and not need anywhere near the tokens.

Thats the thing - these models see and predict tokens. For any real engineering you get more bang for your buck using math.

As someone who has been working on this space for a while (not affiliated with govbase) this is really hard. Between eliminating the sycophancy that seems baked into LLMs and dealing with generalized hallucinations - it's freaking hard. I spent this weekend trying to figure out how to get my system to stop telling me the SAVE Act would be fine because it doesn't say what the process for if birth certificate doesn't match current id.

No, I haven't found a good solution yet - I'm going down a rabbit hole of basically crawling the entire federal register for referenced legislation and then adding in an adversarial agent to see if that can spot gaps.

Would be cool to see some logs on server load, it's one thing to claim "scale to thousands of requests per second" it's another to show the logs of a €4 VPS doing it. I don't doubt it, when I did a similar experiment with golang I never noticed any real issues (aside from a misconfiguration at the proxy requiring a min size on the SSE response) but it would still be neat to see the logs/active users.

SQLite via libsql is my go to for lightweight stuff.

Has vector embedding columns out of the box.

Can you add repos after starting the container? What about persisting indexes across restarts?

Still, neat. Glad to have an easy to deploy open source tool like this.

The number of times I've looked at the speed limit my car thinks it is supposed to be and seen 80 on what is effectively a residential road...

Or that the highway near my house that the car seems to think is 35.

Barring also adding some sort of managed database owned by the state I don't see this being remotely viable.

Squarespace.

Different company than Square.

So https://github.com/BerkeleyHCI/PolymorphicBlocks?tab=readme-... may be of interest to you.

It's ok, I did play with it a bit but the tooling was a bit of a pain to get set up. Not going to be as polished but it's not $1000/mo either.

(It looks like JITX is handling layout and PB is dealing with the netlist side, but it's in the same space)

Maybe if the only thing you're reacting to is other vehicles or the road. The number of times I have to slam on my brakes on that particular road because of animals running into the road is way too high. And no, not always deer. I've come around curves and just had someone's dog sitting in the middle of the road on multiple occasions because for some reason people think it's totally safe to just let their dogs roam.

It’s also difficult to determine if someone is speeding from data.

For example the road I live off of according to the speed limit the car thinks goes from 40 to 65 to 25 to 65 to 40 in about a 4 mile span. Spoiler it does not. It is 40 the whole way. But according to the car I am either going 25 under, 15 over, or exactly the right speed.

(And the 65 section in the middle? Blind corner. Idk where it’s getting its data but it is very very wrong)

No, I actually like that approach better than the traditional ECS approach. Though I assume it both adds and removes some complexity. In a traditional ECS approach you can depend on certain systems running in certain orders because you depend on them having run. Then you refactor, forget this dependency, and suddenly you have changed behaviors and you have no idea why until you remember this. If, as I assume, you have per step immutability that removes a whole class of bugs at the cost of a slight increase in complexity.

Your ECS API looks very similar to that of Bevy's, were you inspired at all by that?

From the little I can see of it, it looks like you don't allow writes to data and instead prefer to return a new thing based on a query? Based entirely on the below.

``` @system def control( sensor: Query[WorldPos, WorldVel, Goal], rw: Query[RWAxis, RWForce] ) -> Query[RWForce]: ```

I assume this is due to the distributed nature of your simulation framework?

Either way, neat. Very interested in this though for more terrestrial applications.

Yes and no, for text type stuff? Yes you're right. But I think in the vision space synthetic data will remain useful for a lot of things. I'm currently working on building a pipeline for personal projects to go from CAD models of environment to segmented training data. So far it looks almost as useful as real world data at a fraction of the cost of manual labeling.

Likely stainless steel. That’s what they made other tools for working with it out of https://youtu.be/myEFGcyInIQ?si=HmJS2kma_skWyKJ8

And tools that facilitate and encourage that behavior.

I’ve tried to do this but the defaults on GH means even if I stack 5 commits each doing atomic things nobody reviews it as anything other than a single patch. And if I split them into distinct PRs the last one looks like the giant PR. And we’re in now I’m shepherding 5 PRs.

Until we have tools or at the very least processes that encourage that behavior nobody will do it.

And don’t get me started on commit message display.

“Start accepting”

Huh. Weird. I don’t really even hide that I’m always working towards my own goals. This includes keeping an eye out for opportunities that better align with them. And I’m a firm believer that having a discussion and seeing what options there are is healthy and good. There have been many times in my career I’ve spoken to recruiters and come away with a better appreciation for where I was.

Didn’t realize this was not a normal approach.

Oh, I had to drop out of the process last year and have since found somewhere I'm actually really enjoying. I mostly wanted to follow up so if someone in my network asked me I could give them the most up to date information possible. I'd feel really bad if I was giving out of date information. I'm glad you've updated your process.

Not mentioned in the emails I had. They did suggest splitting it into 5 and 3 hour chunks but I couldn't fit it into my schedule (between work and my other half's health issues, I had to move around a lot due to several days of ER visits)

First off, congrats.

Second, have you guys had a chance to revise your hiring pipeline? When I spoke to your recruitment team a bit over a year ago they were asking for 8 straight hours on a zoom call as part of the software engineering interview. I had to drop out of the process because fitting that in was exceptionally difficult. (They did offer to split it into 5 and 3 but I still couldn't fit that in). They remarked they were looking to update that process a bit.

The above aside - I remember thinking everyone I spoke to there was pleasant and enjoyed working there. I was sad to have to drop out.

Best of luck moving forward.

Thanks, I've got one of their depth cameras that's been ok. I didn't realize they'd expanded their line so much. Glad to hear about the API/SDK improving, last time I mucked with it a year or so ago it seemed like it was underdeveloped.

Going to have to dig into the sensors they use - had passable luck with non ML tasks using dirt cheap camera modules from laptops running at low resolutions right up until I started moving the cameras at all and then it became a blurry mess because they were so small their exposure times were high. (I'm trying to also avoid having to put a bunch of illumination near the cameras so it doesn't entirely look like a biblically accurate angel)

What should we run it on if we want to deploy it on mobile systems with limited access to network and extremely restricted power budgets?

For a hobby project I'm trying to find a solution - Power budget for multiple is sub 200w. Need to run inference on a lower resolution video stream (or multiple, that be nice) to do object detection. Cost is a factor because I need to have multiple angles to determine where in relation to a mobile platform the target object is. I'm looking at the Coral.ai board because RPi like boards lack the ability to do ML tasks at reasonable FPS and NVidia seems to have abandoned the lower cost side of the market since the Jetson Nanos seem to be less and less available. (Not that Coral.ai boards are available at all...)

Having left a company that also rhymed with Frugal - this somewhat checks out. I do think smaller companies have more options for people to remain ICs somewhat longer. The catch is that ICs do somewhat more than write code.

A few months ago I was the engineer supporting a new ask from marketing, after listening to their requests and evaluating what tools we had I wrote exactly 0 lines of code and we launched on time and without issues in part because I figured out how to use the levers we'd build into existing systems to do what they needed. It's engineering work but not development work.

Outside FAANGs there's more focus on contributing to success than simply contributing LOC. (I left where I was in part because I felt I was penalized for finding efficient ways to do things rather than churning out LoC)

About 40 minutes outside Little Rock.

They are also a great place to cut your teeth on the FAANG type problems because, as it turns out, they face similar scale issues.

There's a lot of "second tier" tech companies that have some really cool problems at massive scales that also pay really well. Not as obviously sexy as FAANGs though.

I mean, as someone who just made almost that exact move... (and not even into the relatively blue by comparison Bentonville area) I think more people might be willing to make the move than you'd think. I mean, it's probably still under 50%. But that area of AR isn't too bad.

Leaning on the stereotypes - I regularly saw more confederate flags and 'lets go brandon' type stuff in the Portland metro area than I have in AR (and I've spent a fair bit of time here) and I've been met with minimal grief here in spite of being the weird dude in the area with an electric car and shoulder length purple hair. Yes, Harrison still has their billboards that are pro white pride radio but there's other places in the state.

I'm a full time remote person but my move here was prompted by a combination of finding a place here that met our criteria and being nearer to my other half's family.

So, some of this also depends on the implications of the word caregiver. I went through a year in which I was the sole care giver to someone that needed assistance walking most days and some days even with their walker would fall. (Fortunately for us, this turned out to be temporary and they are doing much better now, still some lingering issues...)

If you are providing larges amounts of care daily? Yeah, it's brutal. Between work and helping them I was maybe sleeping 4-6 hours a night and regularly having to deal with emergencies. I don't have a great answer in this case other than lean on family. We couldn't do this, but I know folks who are in similar boats who have had some success here.

If it is more a slow illness, this is the right time to find something stable. I know it's tempting to want to find the eye popping salaries but finding something that covers costs reliably and you can plan is much easier. Stable may not be glamorous but it keeps insurance paid for.

Depends, I left Google about half a year back. One of the roles I was considering was a former employer, a non-profit, which would have been roughly a 50% compensation cut. But I loved the work, and loved the team. Ultimately I went another direction because of some poor choices they made around remote work.

The discussion that led to it was "I'd rather we have less money than you be miserable all time time" from my other half. The culture at Google isn't for everyone. I didn't fit with it (or at least in my org) and it was a constant source of frustration on both sides.

So, I'm not diagnosed (though I recently had a member of my care team recommend going through an evaluation to maybe shed some light on things and am starting that process). Truth the "ask your GP" is generally the answer to any question on how to start a process in the US medical system.

(US Based Answer) - Talk to your GP and they can likely recommend a local specialist who does the testing.