Regard | Senior Engineers | NYC, SF, LA | Full-time

Regard is a New York City–based AI health-tech company building the next generation of clinical insights software. Our platform uses machine learning and large-scale data to help clinicians synthesize patient history, recommend diagnoses, and automate documentation. Series B funding.

I'm Justin, the CTO. We're looking for engineers able to leverage current and future AI technologies for product development and platform enhancement. Generative AI experience is required.

Stack: Python, ts, k8s, AWS, postgres Contact: Email me at justin.davis+hn2602@regard.com with your resume and link to a project you've built.

Correct, this is the actual headline too. 100 tflops sure seems like it'd be more than that, but here we are.

If the headline was "FB8 is ~7% faster when kernel name has 'cutlass' in it...", it wouldn't seem sensational.

I think the interesting part is that it improves performance measurably at all, not the actual number. These people are trying to hit 90+% MFU (though most don't reach it) so this does actually translate to many millions of dollars for them.

Flying is incredibly less expensive largely due to deregulation, and Budweiser has always been brewed with rice. So, yes, this is winning.

The deregulation, more specifically, was repealing Prohibitionist policies that kept microbreweries from operating. Now we have microbreweries everywhere, offering superior alternatives to Budweiser, some of which have grown to be national level competitors.

It is, and small teams were often rebelling against incumbents, so rejecting the aesthetics of big companies was a good way to attract talent.

Seriously. It’s not like that was the actual purpose of html or anything.

Hadn't thought about it this way before, but given that LLMs are auto regressive (use their own data for next data), they're sensitive to error drift in ways that are rather similar to analog computers.

Classic spinal tap. “These go to eleven.”

Only if it works >4% of the time.

Only if you intend to run the scam only once, or if all of the work is completely bespoke and not reusable for future attacks.

That seems unlikely. I'm pretty sure there's actually a lot of economies of scale here, where the attackers' pipelines will become vastly more efficient and higher quality over time, with each attack requiring less manual work.

Well a carbon tax would essentially end coal.

Or perhaps you overestimate how many people hold any view of Boeing at all.

They're fairly well known in europe after the MAX debacle, I don't know how it is stateside though.

or even saw the story. Tech nerds tend to pay a lot more attention to the news than most by a large margin in my experience (other than people that are actually in the journalism business)