Pashi (YC W20) | REMOTE / US / India | Full-time | Compiler engineer | https://pashi.com
Pashi is building the operating system for manufacturing. We already work with some of the largest manufacturers in the world, and our software is being brought to bear against some of the thorniest problems that plague the most mission-critical high-volume production processes ever constructed.
We're currently a very small, very technical team spread across India, Europe, and the US, and we're looking for folks with strong backgrounds in programming languages and compilers to help us build better programming abstractions for physical processes.
Email [email protected] with a paragraph-length description of any relevant experience if you're interested.
This has been obvious for months -- Romer has been yelling from the rooftops about it -- and yet no major western govt save the Brits have attempted to even try this strategy. Madness.
As it turns out, you can build most distributed protocols on top of an API which only allows you to append to and read from a shared log -- for example, if you want to do consensus across a number of actors, you can just have each one write their proposed value and the current epoch to the log, and the value that is written into the earliest log slot wins. In general, shared total orderings are equivalent to consensus -- this is clear when you consider that State Machine Replication (Paxos, Raft) is all about ensuring that all actors perform the same set of commands in the same order. If you're looking for a more detailed exposition of the shared log abstraction and its uses, you might benefit from reading the Corfu (https://www.cs.yale.edu/homes/mahesh/papers/corfumain-final....), Tango (http://www.cs.cornell.edu/~taozou/sosp13/tangososp.pdf), and FuzzyLog (https://www.usenix.org/system/files/osdi18-lockerman.pdf) papers from the lead author that precede this one.
Netflix gained ground in original television by funding bold new stuff that nobody else would (Orange Is The New Black, BoJack). They've now becomes slaves to their data despite the fact that the signals they rely on are hopelessly confounded. It's possible that they've hit upon the right strategy, but I both suspect and hope that isn't true.
Starting an independent project or two in an area you haven't played in before might be both fun and useful. I have a somewhat long list of ideas I don't have the time to do if you're looking for inspiration -- feel free to email me if you want 'em (soham at soh dot am).
Hi folks, I'm the host of Segfault, which is a podcast about computer science research. In this episode, Cornell Professor and former Facebook AI Researcher Bharath Hariharan joins me to discuss what got him into Computer Vision, how the transition to deep learning has changed the way CV research is conducted, and the still-massive gap between human perception and what machines can do.
Happy to answer any questions about the stuff we talked about, and would appreciate any feedback you might have.
FYI, in addition to the audio, there's a complete transcript available on the page as well.
Sadly we're not likely to support an RSS feed for Today in Indian History for a little while, but if you sign up via email we'll definitely let you know when we add one.
If you want to target ads, which is where the majority of the money in 'data' comes from, personal data about rich people that helps target ads to them is a lot more valuable than personal data about poor people.
In a weird way, this would be a regressive policy -- the data of the richest people is worth orders of magnitude more, so this would just make the rich, well, richer. Increasing corporate tax rates would help the poor much more.