kozikow

26 ms·

241.
by kozikow 9y ago
> How do I know this? Because it's the only logical thing. As a somewhat libertarian person, I will give you my opinion why I think that low taxes are better. As history shows, all known forms of government eventually become very in
242.
by kozikow 9y ago
I was in a group of people that didn't get the H1B in the lottery for a big tech company. Most of us went to work in the European office. Later, about 30-50% of those people went to USA on L1, much closer to "indentured servitude&
243.
by kozikow 9y ago
> In practice, though, H1-B becomes indentured servitude, since they cannot freely move between companies. You can change company while on the H1B. Yes, there is some paperwork, but in practice, everyone who is in the USA on the H1B can
244.
by kozikow 9y ago
Maybe the logic is another way around? Companies that have a clear product-market fit and generate enough revenue to self-fund and don't have competition closely behind their backs have much less reason to raise. Maybe you are mistakin
245.
by kozikow 9y ago
Probably ended up on EA PM's presentation justifying current practices. Hate all you want, "pay to win" model works for extracting as much value as you can from each user. Same concept as Tesla artificially reducing mile rang
246.
by kozikow 9y ago
Any idea where? Do you mean the web interface? I can't find it anywhere: https://drive.google.com/file/d/1nEtxRgXs1yD1bV2-aDjkugc7Ocd... .
247.
by kozikow 9y ago
I investigated targeting individual Uber drivers for street view mapping. Problem is that Uber drivers circle around the same area a lot, so it is not optimal to use professional 360 set. On the other hand, there is no good consumer grade c
248.
by kozikow 9y ago
Shameless plug: our company (tensorflight.com) works on extracting building information from a few sources including street view. E.g. we extract a number of stories building have based on street view image. Please get in touch at kozikow@t
249.
by kozikow 9y ago
There are quite a lot of 360 consumer cameras you can just plug into your cell phone. Quality is still so-so, but it's getting better.
250.
by kozikow 9y ago
How do you get access to historical here street view imagery?
251.
by kozikow 9y ago
Google also have oblique (45 degree) anywhere you have a 3d model. For some reason they do not show it in google maps and you need a separate API to see it: https://developers.google.com/maps/documentation/javascri
252.
by kozikow 9y ago
It would be nice if I could buy it on Amazon.
253.
by kozikow 9y ago
Adding features too quickly was one of problems with swarm.
254.
by kozikow 9y ago
> Does Python have a library that works with the GPU for image processing? tensorflow. It's not just deep learning library and you can implement any tensor operations there. 99% image processing is just tensor operations.
255.
by kozikow 9y ago
Cryonics + post enough incentive for someone to wake you up. The best incentive for someone to wake you up I came up with is putting lots of money in cryptocurrency and remembering the private key and writing "Will pay X BTC to anyone
256.
by kozikow 9y ago
I use minikube for dev and it's pretty good. My description of my minikube+pycharm setup: https://kozikow.com/2016/09/16/using-pycharm-docker-integrat... . The only problem I didn't solve yet is deb
257.
by kozikow 9y ago
> or a continuing push to marginalize docker by the Kubernetes folks. It's not like kubernetes folks very pushing to marginalize docker for bad reasons. docker runtime was one of the bottlenecks of kubernetes in production and docke
258.
by kozikow 9y ago
The top captures all the economic growth: http://time.com/money/4371332/income-inequality-recession/ .
259.
by kozikow 9y ago
Who needs anything else except deep learning? \s
260.
by kozikow 9y ago
I am using org mode daily. The biggest problem is onboarding other people on the team - google docs are still easier for collaboration. Many times I wrote an initial version of the doc in org-mode and exported it to google docs for feedback
261.
by kozikow 9y ago
Probably incorporation affects result by a wide margin. E.g. our company (not YC alumni) is incorporated in Delaware and based in Mountain View, but both founders are Polish and half of the employees are in Poland. On such list, we would co
262.
by kozikow 9y ago
> Governments do however have one source of leverage over crypto networks: exchanges in which users trade fiat currency for crypto. Not really: - localbitcoins or localmonero, if you want to stay super secure you can use an option of &qu
263.
by kozikow 9y ago
Different anecdote: I am not regular whole foods, but went out of curiosity whether there will be any event related to amazon buyout or just check out if there any significant price drops.
264.
by kozikow 9y ago
Not necessary pre-relief, but some companies assess risk of a damage prior to the disaster. E.g. at tensorflight we are assessing building insurance risk via computer vision and significant part of features we detect are useful for hurrican
265.
by kozikow 9y ago
I listen to python podcasts: https://talkpython.fm/ and https://www.podcastinit.com/ . Outside of purely technical subjects, I also like "talking machines" about ML: http://www.thetalki
266.
by kozikow 9y ago
We already got some post-Harvey drone imagery, but not much yet - government restricted civilian drones. So far, we primary have been annotating imagery from prior to catastrophe or other catastrophes.
267.
by kozikow 9y ago
Update: We also started annotating humans and animals.
268.
by kozikow 9y ago
At tensorflight.com we are building deep learning model for hurricane Harvey damage segmentation. We already have 12K instances of partial building damage and 400 of total damage annotated and preliminary segmentation model is just being tr
269.
by kozikow 9y ago
Tensorflight | Mountain View, California or Warsaw, Poland | Onsite (preferably) or Remote | Fulltime Hiring enterprise sales and business development, computer vision and engineering. At tensorflight we are creating computer vision and mac
270.
by kozikow 9y ago
It's hard to find an appropriate definition. What if AI is there theoretically only to provide suggestions to a human operator, but in practice, the operator gets so used to AI's predictions that he always goes with the AI's

More ›