lmeyerov

69 ms·

841.
by lmeyerov 4y ago
Yep, referring just to CS. Different even from other STEM fields. In CS, stipends are below median income, but get close to median when students do the occasional R&D internship. The main CS grad students I've seen struggle are tho
842.
by lmeyerov 4y ago
Yep. Something like a masters is different, which explicitly is about not continuing in academic research. In contrast, the number of folks going on to a phd with one of their potential outcomes being a professorship is probably closer to 9
843.
by lmeyerov 4y ago
Agreed with problem, not sure about solution Ex: 10X smaller enrollment. In top programs, most competitive undergraduate applicants have _already_ published top scientific papers. Making 10X more competitive will turn up the pressure even m
844.
by lmeyerov 4y ago
Yes, weaker programs already do so. The faculty generally are top notch, and superstar students do seek out individual professors due to whatever alignment and succeed. So IMO it's more about the student, there are definitely places wi
845.
by lmeyerov 4y ago
He'd be bankrupt many times over without his ability to work massive $ from the US Gov - I'd argue it's one of his biggest strengths, trailing somewhere after his stock market influence abilities, both of which seem key to ho
846.
by lmeyerov 4y ago
Numbers-wise, something like 3% of phds lead to professorships: it's not quite as competitive as becoming a professional athlete, but not far. Peak competition. Most professors watching their interview candidates nowadays freely admit
847.
by lmeyerov 4y ago
Yep :) Unfortunately, guns is a US problem, not a SF one. Within that, most violence is between people who know each other, so that's a funny one to work through. We moved our Oakland office to SF not long after watching a drive-by sho
848.
by lmeyerov 4y ago
I've lived in the same spot in the mission for 10+ years (and been around longer), and other folks in the building for 20+: the mission hasn't regressed to the gang wars era (we have stories there...), but between the break-ins, h
849.
by lmeyerov 4y ago
Just spent an hour this morning helping a customer's 2 js devs do a hello world setup because of the first one. And even as someone experienced here, it still took me 10-15min to untangle the core issues. (In this case, enabling esm as
850.
by lmeyerov 4y ago
Until you want value-based fraud prevention layers like insurance. Keeping fixed fee would knock that out of alignment. But perhaps your point would be that should be some sort of opt-in higher SLA tier so the transaction layer fee is low b
851.
by lmeyerov 4y ago
Less is more, especially later: - Career-wise*, work with good people on something important, maybe switch it up every 3-5 years. That is under your control and steady. Professionally, daily wins/losses, keeping up with joneses, etc. d
852.
by lmeyerov 4y ago
I thought the SEC filings both have a number and a risk of being wrong? Re:parroting, while that seems to have been a pejorative, I wasn't actually copying from anywhere, just a smb executive sensitive to this sort of thing during m&am
853.
by lmeyerov 4y ago
If they lied on SEC filings about this, he bought something not as advertised, so I'm not so sure...
854.
by lmeyerov 4y ago
What is a few clickthroughs for the owner of a home box or regular co box can be weeks for teams or even a deal breaker in critical envs that are locked down. The money people know where the money comes from when making strategic decisions
855.
by lmeyerov 4y ago
Yep. We got burned out on the lies RHEL8 pushed to serious orgs (utilities, gov agencies, etc.) around podman readiness & replaceability over docker. I'm glad people are getting their bonuses and good for docker to have competition
856.
by lmeyerov 4y ago
Maybe put differently: You can be 9-5 FAANG type and retire as early as 40. In contrast, due to the delayed paycheck for specialists... they're looking at another 10 years after that. Likewise, at FAANG, you're living large as a 2
857.
by lmeyerov 4y ago
Latest pygraphistry has this flow for free and OSS, just `pip install graphistry[umap-learn]` or, for transformers, `pip install graphistry[ai]` :) And per the article, with pluggable sentence transformers -> UMAP automatically as part o
858.
by lmeyerov 4y ago
Useful to think through the numbers. Few American dr's make $1M. Most of those that make many multiples more than nurses are specialists, and they are still much less than $1M. I do know ones that do better... Because they are involved
859.
by lmeyerov 4y ago
Selective memory or maybe from a place of privilege? His team called their initial orders their 'shock & awe' strategy and resulted in colleagues not able to come to school+work because they were not allowed back to the US aft
860.
by lmeyerov 4y ago
I love this: - 'Distribution' is really the hot button this is about, even if I didn't see the word there. "Build it and they will come" is rare. Such product/market fit thinking sweeps that difficulty under th
861.
by lmeyerov 4y ago
Depends -- As a remote team, once we have someone, we don't want to lose them: we are mutually invested. Likewise, overlapping timezones help, esp for our more R&D areas. So if you are good, you are good. At the same time, we are t
862.
by lmeyerov 4y ago
Yes! I'm a fan & practitioner of NVC, mindfulness, careful work life balance, etc for being able to treat startups as a marathon vs sprint... But I've observed people overuse & abuse these tools to rationalize prioritizing
863.
by lmeyerov 5y ago
Super cool, and a lot of contributors! Can this work the rest of the wireguard ecosystem (agents, UIs, ...) for a full VPN soln without involving the VC-tied company?
864.
by lmeyerov 5y ago
Yep we had it rejected w an enterprise we work with as the org needed to own the full control plane so we couldn't bring it in, and not on the schedule for the org's security team for them to bring it in. Making a smarter, easier,
865.
by lmeyerov 5y ago
Yep, dask_sql is probably a better comparison here (and similarly arrow-friendly). We've been having pretty good experiences here, though clearly young. For in-process, we're sticking with pandas/cudf (GPU dataframes), and li
866.
by lmeyerov 5y ago
Each issue can be from a 1 line fix to a week or even 2 week thing. But by that point, probably a metaissue that corresponds to other issues or issue with inline subtasks (GitHub has check boxes). Features cut across code, so no 1-1 mapping
867.
by lmeyerov 5y ago
Is there a clean way to do reuse this for multistage builds? ``` FROM nixery.dev/shell/git/node14/python3.8 as debug_extras FROM our/production:1.2.3 COPY --from=debug_extras /nixstuff /ubuntu/stuff R
868.
by lmeyerov 5y ago
having firecracker use a local cache vs docker with a remote cache means you'd be benchmarking network hardware & bandwidth, not docker vs firecracker
869.
by lmeyerov 5y ago
Yep, buildkit does incremental builds quite well We find the dominating factor in (our) incremental builds / CI to be network/io caching, which has less to do with firecracker/docker and more with the surrounding hw/sw (
870.
by lmeyerov 5y ago
Docker largely entered Enterprise (regulated & security conscious markets) like utilities, us gov, banks, etc. via rhel 7 / centos 7 . A lot of people doing a lot of compliance work everywhere, from approvals to infra to audits, in

More ›