lmeyerov

55 ms·

721.
by lmeyerov 4y ago
Ex: With all the Kanye meltdown stuff going on, which is "general interest", Twitter has been amplifying the tiny % of antisemitic voices in the black community and mixing them into my feed. That is quite distracting for a site I
722.
by lmeyerov 4y ago
That's why we built Graphistry :) Can be used code-free (drop in a csv or log file) or via analysis workflows like from Jupyter notebooks and internal dashboards. We are roughly parity for Gephi's features, and pushing (far) ahead
723.
by lmeyerov 4y ago
(Big fan of graphviz, it was such a shame when AT&T pulled funding!) Maybe helpful, more for visual graph data analysis than say arch diagramming: we build Graphistry specifically for interactively analyzing more data than these tools,
724.
by lmeyerov 4y ago
We are in a similar boat from the b2b side - we sell a small % of our software in the Amazon and Azure Marketplaces. Importantly, we don't compete in the Marketplaces for advertising, just work with them for final purchasing, and what
725.
by lmeyerov 4y ago
FWIW... None of those are marketing investments into evergreen communities that will keep growing the app. Build channels, not one-offs. Exceptions exist, like social networks where you can do stunts until it grows itself, or ads when you h
726.
by lmeyerov 4y ago
agreed, reading this article was confusing, the python baseline is far from our reality for reference, we're aiming for 1-100 GB / second, per server, in our python etl+ml+viz pipelines interestingly, duckdb+polars are nice for sm
727.
by lmeyerov 4y ago
The CHIP act makes way more sense to me financially + getting the bill passed given this stuff. Long-term this does just push China into self-sufficiency faster... But near term, US co's profit? I was mostly curious about the jobs mark
728.
by lmeyerov 4y ago
I've seen 10+ of these advertised this year despite not being in the space -- is there some shift that enabled them, like open pydata connectors 5 years ago and now enough steam that several have grown?
729.
by lmeyerov 4y ago
Yeah more of a OSS network traffic inspector for security monitoring. I think it ships with IDS rules nowadays, but we see it generally more for manual impl. We see often fed into Splunk or ELK to feed custom detections, and especially enri
730.
by lmeyerov 4y ago
Kind of the reverse, it's an interesting feature people haven't been using yet / colab can't do well yet: inheriting coding style a MAJOR element of prompt engineering in stable diffusion for image generation is specifyi
731.
by lmeyerov 4y ago
Good point -- Yes, no guarantee No, strong ugrad academic perf is not required. Professors review applicants, you need to identify and win a champion based on your aligned research interest & progress. Eg, research-y side projects and e
732.
by lmeyerov 4y ago
Trick: apply for a PhD and they will pay you to get the masters, and you can drop out at any time And agreed with others. If you did not do a great upgrad / did not sufficiently self-teach, and you want to quickly level up on tech fund
733.
by lmeyerov 4y ago
We mostly want this for development of CUDA apps, not even super worried about performance. Increasingly, I'm thinking just a mobile eGPU Ubuntu node you plugin and ssh into, and just a nice remote setup...
734.
by lmeyerov 4y ago
The economic value isn't measured in revenue but the stock price, which you get paid your bigger bucks in. Their stock dropped 86% over the last year, so market value per engineer is not the same. Compensation was made with that in mi
735.
by lmeyerov 4y ago
Stimulating question! Maybe some useful observations as we do client webl <> server cuda acceleration of vector data (our use case is big visual graph analytics) and have to handle large + real-time contexts: * SVG at those levels so
736.
by lmeyerov 4y ago
I wonder if there's an intuitive ratio here, like 10 freezes for every layoff, and 10 slowdowns for every freeze Freezes and slowdowns don't really need to get announced. We've definitely seen an uptick in candidate quality (
737.
by lmeyerov 4y ago
We got stuck on the Nvidia opencl drivers before we went deep into k8s networking Our experience with local k8s dev is the issues are more on the k8s distro side. So at least for local dev, I'm guessing the happier path is sticking wit
738.
by lmeyerov 4y ago
We burned an evening just last week tracing an Intel wifi driver issue to missing kernel headers that required upgrading the kernel to a new, non-LTS version. And only then did we move on to Nvidia drivers. So no, still not the year of Li
739.
by lmeyerov 4y ago
The likely goal is decreasing the bite out of 24/7 on-call wrt customer responsiveness and service uptime Technical user support & service uptime: can simplify by design, docs, self-healing. For a focused service, remaining timely
740.
by lmeyerov 4y ago
Automate Simplify business + product design and make arch boring Prioritize getting more revenue so you can hire
741.
by lmeyerov 4y ago
There's a funny bite here that seems long-term good. The gov has a large culture of tapping integrators who do not give back to OSS, just use, basically the middle man, and leave behind fragile one-offs. Such abandonware should overwhe
742.
by lmeyerov 4y ago
I'd say that's a weak test of capacity. Would love to see on Azure - T4s or an equiv aren't even really provided anymore! We find reliability a diff story. Eg, our main source of downtime on Azure is they restart (live migrat
743.
by lmeyerov 4y ago
(one of the authors) Yes! We carefully distinguish vector search vs vector databases throughout the article. The end sections on architectures touch on that, such as questioning compute-tier solution vs not. Reality vs marketing seem to div
744.
Measuring the popularity of different vector databases(gradientflow.com)
63 points by lmeyerov 4y ago|18 comments
745.
by lmeyerov 4y ago
Can be good or bad * Prioritizes ownership, working hard (not just smart), and solving the problem, even if it requires manual effort (not beneath you, grinding, ..), skills you don't have, emotional rollercoaster scenarios of startup
746.
by lmeyerov 4y ago
agreed, and we've heard heart-attack-level pricing pushback for the few db's that do scale well likewise, for analytics, we're seeing a bit of a split: - DB side: Traditional real-time graph analytics (pattern search, ...) ca
747.
by lmeyerov 4y ago
Thanks for kind words all :) Scale - backend: As we're rapids.ai-native (helped start early days of both Apache Arrow + Nvidia RAPIDS.ai), we work with customers doing billion-level nodes/edges in interactive time on GPU servers.
748.
by lmeyerov 4y ago
Good reading, I've been curious! Some observations as a team active here. We're adjacent / complementary because we make graph db's (and regular SQL/databricks/etc.) more actionable via rich & scaled graph
749.
by lmeyerov 4y ago
We have tried + investigated a variety, trying to avoid unnecessary shade for indiv co's. Hundreds of millions maybe even billions have gone into funding this space now, so folks get twitchy :)
750.
by lmeyerov 4y ago
We've been looking at a few for what I thought "basic" AI data prep tasks like scraping with backfill & periodic refreshes (basically a bunch of REST queries with care not to overwhelm the targets; 2000's era backpre

More ›