kekeblom

Email me at hello@keke.dev. Lets figure it out.

I just built an app for this, if you have a LiDAR iOS device. https://keke.dev/blog/2021/03/10/Stray-Scanner.html

The iPhone costs more than $500, but you can also use if for other stuff.

Qualcomm recently released a platform which has 5G and decent compute, which could be used for this. https://www.qualcomm.com/products/qualcomm-robotics-rb5-plat...

Likely would have to build some autonomy into the system to handle scenarios where the communication gets patchy or the user is about to crash into something.

Now you only need to install 5G on the Grand Canyon.

Well there is a way to upgrade. They even provided an automated tool that would convert code and warn you about the handful of changes.

Likely a key factor is leaving support for the old version. If they had immediately deprecated python 2, users would have quickly updated their packages over to a supported version.

Scott Kuindersma shared some high level details of how they do some of the stuff they have done in their demos in this talk https://youtu.be/EGABAx52GKI

He doesn’t say exactly how they have done things on the low-level, but there is enough to go out digging into what he and other people have published.

I had an issue recently where the form contents filled and saved with Preview.app would not show up in acrobat reader. I've encountered this in two cases so far, with two completely different documents.

As for pick_up_object() startups, there are:

- https://picknik.ai/

- https://covariant.ai/

- https://www.righthandrobotics.com/

- https://lyro.io/

- https://www.universallogic.com/

and probably others.

What I think is driving this is the lack of competition. No-one is going to take on the Kindle. Not because they don’t think they couldn’t make a better device, but because the don’t think they could create a digital bookstore with enough titles to be useful.

Right, but there's still a lot of overlap between features. Take layout for example. If you wanted to arrange boxes using html and css, you have quite a few choices. You could use at least: tables, floats, flexbox or css grid. Imagine if there was only one of those that would cover all possible layouts.

The web (html/css/javascript) standards have become so wide and complex that it's very hard to build a browser from scratch. It gets worse by the day, as more features get added to those standards.

This makes me wonder if you could define a subset of those standards that would support 90%+ of the use cases, would be simple and efficient to implement and would be supported by existing browsers. You could then push this light version of the web as a new base for a future standard.

The problem would of course be that you would have to keep around a different browser to render old sites. Could still be worth thinking about. Doesn't seem feasible as a long term strategy to have browsers support thousands of apis and increasing.

C++ through CUDA is by far the most popular option. There is some support in other languages but the support and ecosystem is far from what exists for CUDA and c++.

I ported a particle simulation I built for a course to use CUDA https://github.com/kekeblom/mpm

Mainly to gain more first-hand GPGPU programming experience.

Why don’t these things just generate svg and serve that embedded in the html?

The city of Zurich has what to my understanding is public fiber which is available to most buildings. (I assume there are some exceptions). It is managed by the electricity company run by the city. https://zuerinet.ch

Setting up routers, billing etc. has been delegated to any third party ISP willing to do the job. I think they can charge what they want, as there are some very small differences in price. I pay ~44CHF/month for 100mb/s. While not dirt cheap, It's ok considering the price of other stuff here. Service so far has been very reliable.

What boggles me is why modern buildings still have so many manual doors with handles. At least in the building I work in, you could easily eliminate a ton of doors and not reduce access control flexibility. The rest could be made automatic with some keycard authentication.

You have to press shift in order to unlock more power and clear that jump.

I have a Lenovo P52 at work. Compared to a Macbook it's not that great. You can't open the lid one-handed as the bottom sticks to the top. The build quality is super flimsy. The touchpad is not nearly as sensitive and accurate as on a MBP.

I run Ubuntu on it and currently my Nvidia drivers are not loading correctly so it's running on the intel integrated graphics. When they were, all GUI applications would crash when waking up from sleep. Will have to look into it and figure out if I can find a stable configuration.

The thing that actually sucks is in python is parallelism.

To me it seems weird that the law should even consider the gender of a a person. I do realize most countries have different legislation for men and women which is just very hard for me to understand the justification for.

Can someone think of corner cases where it should matter whether a person is a man or a woman?

It’s actually fire safety code in Finland that doors have to open outwards. It should be easier to evacuate the building in case of a panic.

You might still find old buildings where this is not true and they probably just predate the regulation.

The transaction fees would end up going to someone (e.g. miners but some implementations might have some other system in place) and eventually they would end up spending them. At least that is how current cryptocurrencies work.

The problem is that most people making the decisions own one or more units. If they decided to build more housing, the value of their investments would decline.

Location: Helsinki, Finland

Remote: Possible

Willing to relocate: Yes

Technologies: Python, Scala, TensorFlow, PyTorch, SQL, Rust, C++, Javascript, Node, Ruby, Stan, Theano, Git, Linux, Redis, Matlab

Résumé/CV: https://kenken.keybase.pub/resume.pdf

Linkedin: https://www.linkedin.com/in/kennethblomqvist/

Github: github.com/kekeblom

Email: kekeblom@gmail.com

I study machine learning and data mining at Aalto University in Helsinki, Finland. I'm looking for a master's thesis project within machine learning. In our school, the thesis project is a 6 month long research and/or engineering project which can be completed at a company or at university.

I study mathematics as my minor. I have several years of experience as a software engineer.

Feel free to contact me!

Considering that an ico in oculus’s case would mean that they would issue a token which could be exchanged against a device, I fail to see how the token could increase in value in any significant way.

There is a project called Basecoin. Here is an article about it: https://www.coindesk.com/basecoin-revealed-a16z-metastable-s...

Unfortunately, I don't know much more about that project at this point.

It seems like it is.

According [1] a 1080 Ti can produce about 675 hashes a second for Zcash. Using [2] we can see that one card would net him about $2.5 per day. For six cards we get ~$450 per month which is enough to pay his electricity bill and probably internet costs.

The graphic cards are quite costly and the payback time for the card thus pretty long.

[1] http://zcashbenchmarks.info/ [2] https://www.cryptocompare.com/mining/calculator/zec?HashingP...

Practical issues aside and from an individual’s perspective, why use an inflationary currency at all if you can use a deflationary one?

Fyi, Google does do business in China. They have offices in Beijing and Shanghai at least. Their phones are most probably also manifactured in China which is pretty big business.

https://en.m.wikipedia.org/wiki/Google_China

Pretty much the same thing has happened in natural language processing. Previously trained linguists would spend lots of time on carefully crafted features. Now you just throw a bidirectional lstm model at the problem and enough training examples and you are close to state of the art.