lmeyerov

@lmeyerov

CEO : 100X investigations w/ GPU visual graph analytics & automation. Invented: GPU dataframes, parallel browser, flapjax, socio-PLT

San Francisco
Vrijeme pridruživanja: srpanj 2008.

Tweetovi

Blokirali ste korisnika/cu @lmeyerov

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @lmeyerov

  1. Prikvačeni tweet
    11. sij 2016.

    I'll just leave this here:

    Poništi
  2. prije 10 sati

    Happening tomorrow. Cool way to learn about precision medicine / genetics / cancer, GPU data frames, and UMAP clustering ("uniform manifold approximation & projection"), chat w/ folks in the middle of it, and post-show, even try for yourself!

    Poništi
  3. proslijedio/la je Tweet
    3. velj

    Session 02 is Tuesday! Join for cancer genomics live-coding as Dr. (Stanford Medicine) and ( ) experiment with visual GPU UMAP for longitudinal analysis over thousands of cancer patients in GENIE / !

    Poništi
  4. proslijedio/la je Tweet
    31. sij
    Poništi
  5. 31. sij

    OH: "Security product [x] model is to push services so their docs are sparse" THANKS, - Frustrated partner trying to make your product not look bad in front of a joint account

    Poništi
  6. 31. sij

    I love that a famous cryptographer said (behind my back) that PL was dead, and now the crypto bubble burst while the data+ML revolution is increasingly sitting on PL/HPC/distsys/etc. :)

    Poništi
  7. 30. sij

    Just added `pigz` for multicore goodness to Graphistry's auto-backup & migrate script. Just two lines of code and much speedups later.. 😎

    Poništi
  8. 30. sij

    Anyone have a fav Code of Conduct for events? Both for speakers + attendees, and even better if mindful of the heightened needs in the sec/fraud space like around photography? ( cc ! )

    Poništi
  9. 28. sij

    Fun mini-step in our GPU serverless CI: -- Last year: Nightlies & PR updates trigger a docker job that build GPU dockers (multistaged w/ snapshots) & save AMI+docker images+tarballs -- Finally: All the way to prepping our cloud marketplace images. Last stop will be our cloud!

    Poništi
  10. 28. sij

    Maybe a culture diff: Systems people focus on 'weak scaling' for hours/minutes probs HPC people stress 'strong scaling' for compute on a fixed-SIZE prob So: 2009-2020 was weak scaling (google/hadoop+), while 2020 is strong / latency (multi-gpu)

    Prikaži ovu nit
    Poništi
  11. 28. sij

    (this matters for people trying to build systems for automation like ad ranking / fraud detection / ... and interactive analyst use... your big budget project has a clear arch issue that will takes many developer+actual years to work around, and leaves a gap for competitors)

    Prikaži ovu nit
    Poništi
  12. 28. sij

    Found I've been referencing Stuart Cheshire's + amdahl's law a bunch recently. A 2020 corollary of bandwidth vs latency keeps coming up for data dev: "it's easy to make parallel software scale to bigger data, but you can't just layer on hitting < 100ms"

    Prikaži ovu nit
    Poništi
  13. proslijedio/la je Tweet
    27. sij
    Odgovor korisniku/ci

    Makes sense! I was thinking of the scenario where the analyst who receives the alert today was not part of the team that worked an identical case last year. Would be great to be able to identify that a subgraph in today’s data “looks like” that case from our archives.

    Poništi
  14. 24. sij

    cc : maybe another way to explain for the js community is "it started as an interop format for going faster between different data systems and is growing to be an accelerated form of lodash for tables within each lang"

    Prikaži ovu nit
    Poništi
  15. 24. sij

    cc and other web superheroes :) It's getting used at big JS teams like Uber , for a lot of our GPU work in , and even some banks. These APIs should help open up the tent to the wider browser+node community :)

    Prikaži ovu nit
    Poništi
  16. 24. sij

    JS friends: [JS] is an A++ lib when you work with data tables, CSVs, etc.: Accelerated storage & compute, types, streaming, data interop, & more. The high-level data APIs matured, so I wrote up an intro tutorial: cc

    Prikaži ovu nit
    Poništi
  17. 22. sij

    One polling loop removed, one cron added, so calling that a balanced kind of startup day. (Other stuff went well, can't win 'em all!)

    Poništi
  18. 21. sij

    Super excited to announce #02. The GPU notebooks for this episode have been pretty nuts so far and already getting help from the community. Thanks for supporting us on launching, and please reshare the announce 👇 🚀🚀🚀

    Poništi
  19. proslijedio/la je Tweet
    20. sij

    Latest "Read a paper" looks at a paper by : "Empirical analysis of Programming Language Adoption".

    Poništi
  20. 17. sij

    Something I and our users work a lot on. The US economy enables us to force anti-corruption policies on trade partners, such as shell company ownership. This unveils tax evasion, blood money, etc.: the world wins. Years of non-partisan US efforts are getting destroyed.

    Poništi
  21. 17. sij

    (modin on ray. I didn't include dask because, well, see for yourself.)

    Prikaži ovu nit
    Poništi

Čini se da učitavanje traje već neko vrijeme.

Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.

    Možda bi vam se svidjelo i ovo:

    ·