lmeyerov

46 ms·

601.
by lmeyerov 3y ago
Also, the thread title is unintentionally funny: I'd like the interface to be more functional so we can write truly 'point-free' pipelines, especially around areas like memory. Ex: When dealing with multithreading, that makes
602.
by lmeyerov 3y ago
I've definitely got heartache here, and they merit criticism, but it is real We need a lot of pluggability to support diff vendor LLMs and BYO LLMs in Louie.ai, so having langchain has been nice for helping code to interfaces vs vendor
603.
by lmeyerov 3y ago
Not everything should persist, such as in a classic web architecture, analogous to redis vs postgres Ex: For some of our use cases for more-text data sources and APIs we are helping folks index for Louie.ai, we do ingest-time embedding into
604.
by lmeyerov 3y ago
We are seeing similar for Louie.ai . Data catalog yes, semantic no. We are also doing a lot of non-SQL like logs DBs (SIEMs) and graph DBs, which adds even more fun to that kind of question. My hypothesis here is that "semantic layer i
605.
by lmeyerov 3y ago
* agreed on students summer vacation * My own use has lessened but steadied, eg, writing * Coming: A dog-fooding goal for our louie.ai team is to eliminate most of our internal data scientist and SE/SA use of ChatGPT by having more pur
606.
by lmeyerov 3y ago
I appreciate the good faith attempt! https://github.com/graphistry/pygraphistry No 3d/vr/etc unfortunately, we never saw commercial demand in our segments, just looking at bigger datasets and increasingly &#x
607.
by lmeyerov 3y ago
I always liked the Netflix reframing of 'professional sports team's vs 'family' because of that reason. At the same time.. I'm sorry you've had a professional career with so many folks lacking loyalty. I'
608.
by lmeyerov 3y ago
I think that's more true than people, especially pure engineers, would like to think Ultimately companies are made out of individuals going through their own lives and making decisions as they go. Raising capital, selling software, and
609.
by lmeyerov 3y ago
Yes it varies. Some prefer it before jumping in, eg, super senior who mostly do advisory and light consulting. We are a small team so by definition each person gets more ownership than at your average bigco or megaround VC co.
610.
by lmeyerov 3y ago
Thank you, and thankfully we are fine I just feel bad that folks are pretty much disqualifying themselves needlessly. It's a tough job market, and if folks are flubbing it with us that needlessly, I can only imagine what they are going
611.
by lmeyerov 3y ago
I'm not sure where I say we "have some expectation that they [candidates] should be upset and disappointed if they don't make the cut". I would assume they'd be disappointed they didn't get an offer, wasted app
612.
by lmeyerov 3y ago
Yes, we are looking for folks to work with us, not for us. Different mindset & process. We try not to hire ex-FAANG (but occasionally do) in part because of this kind of difference. It's fascinating to see so much resistance to thi
613.
by lmeyerov 3y ago
Agreed all around. For us, folks using our tech and then realizing what it's for is the beginning of much more interesting technical & mission conversations than programming language, monetization model, or RSU vs ISO. I rather ta
614.
by lmeyerov 3y ago
Yes, generally I assume each candidate is having serious conversations with 2-5 others, and non-serious with more. We do the same. There are exceptions, like folks not actively interviewing, but that's the typical case. It is big stake
615.
by lmeyerov 3y ago
Great! I can't fathom having a serious discussion about a 2-5 year career bet, and hopefully even more impacting, and not having a serious look at the actual work or at least the highly related technologies. Some of our best hires have
616.
by lmeyerov 3y ago
Yep. Folks at 200+ and especially FAANG companies are mostly interchangeable. The interviewer is mostly derisking chance of a dud and comparing a 86%er vs 87%er. A few exceptions like at the rarer million dollar comp level. For everyone els
617.
by lmeyerov 3y ago
Graphistry (graphistry.com + louie.ai) | DC | US | Australia | REMOTE | Full-time and Contract-to-hire | Senior data engineering, backend engineering, research engineering, security data science, security engineering Graphistry is a Gartner
618.
by lmeyerov 3y ago
Yep, less about fully shutting off and more about negotiation. This can be Musk doing a direct cut on some of their biggest web traffic to show that they can optimize it down. Or same, but internally, telling internal engineers they only ha
619.
by lmeyerov 3y ago
Apparently their 5-year cloud contract failed to renew today So, there may be a much simpler explanation for why their new rate limits on regular users to can-barely-scroll levels, and how that has all sorts of unintended consequences they
620.
by lmeyerov 3y ago
So maybe 20-30% of dev, which would be overrepresented here?
621.
by lmeyerov 3y ago
I mean, we have found election manipulations like large-scale inauthentic activity of out-of-staters explicitly targeting African Americans, and projects here even to the extent of the perpetrators getting indicted. Other projects were trac
622.
by lmeyerov 3y ago
Yes -- this is a good article on it: https://txt.cohere.com/sentence-word-embeddings As a blackbox, it is a generalization of word2vec to sequence2vec. For example, simply summing or averaging word vectors in a sentence can
623.
by lmeyerov 3y ago
We always spent a bit of time doing extra infra for the RHEL corner of the world to make lives a bit better for IT teams at our banking & gov customers, and the licensing stuff was weird but ok. But when rhel8 came out with heavy market
624.
by lmeyerov 3y ago
Something not obvious to me with these VC diagrams wrt the memory tier being just vector DBs vs also including knowledge graphs Good: We're (of course) doing a lot of these architectures behind-the-scenes for louie.ai and client proje
625.
by lmeyerov 3y ago
Yep! Tech companies over hired for ~10 years wrt efficiency & sustainability, and few want to do layoffs necessary to get 'good' efficiency numbers, instead just close enough that they can maybe reach non-buzzy norms in a few
626.
by lmeyerov 3y ago
At first I was excited -- this is the second time i'm.seeing this advertised, and we are thinking through reliable API call-out strategies for louie.ai -- but then I got confused by the paper: Is this really just tested against 95 API
627.
by lmeyerov 3y ago
Yeah no
628.
by lmeyerov 3y ago
Winner!
629.
by lmeyerov 3y ago
It was a useful read for me, esp seeing numbers on fine-tuning & use. We are piloting a DB analyst tool where users can use natural language to do DB queries, generate AI analyses, make interactive GPU charts, etc, so many nearby questi
630.
by lmeyerov 3y ago
I find it interesting they cast data science as some sort of inherently wrong Other When your profession views math on user & business data as the enemy, and it's for a business...

More ›