deepGem

This is correct. What I meant to say was that in due course, re-training on the fly will become a norm. Even without on the fly re-training we are looking at a small delta.

There's no way you are a loser. 1 successful exit, flaunt it man. No one cares about the exit value. The fact that you had a successful exit is a huge win. I would say that you can certainly strive for a 250-300K salary in the bay area at least.

You should try Walmart for an IC role. Glacial pace of working and a shit ton of bureaucracy, but looks like it might just be the right environment for you now.

Other companies that I can think of along similar lines - JPM, Cisco, Arista or any of the traditional companies.

Your best bet is to network with the hiring managers. I think if you just meet them and speak about your experiences in an event or something you might be able to articulate your experiences better than just in a resume.

Also I have found that drafting your resume and asking ChatGPT or another LLM to refine the wording really works. These LLMs know the ATS or resume scanning keywords and will tailor your resume accordingly.

Good luck

But once it knows it’s pretty certain to become common knowledge almost instantaneously. That’s not possible now. What you learn stays localised to you and may be people 1 degree away from you that’s it.

I don't do plunges but do cold showers. They are not the same but cold showers do have one anecdotal benefit for me, I don't feel cold post the cold shower.

My immunity had been pretty good but am not sure if I can attribute that to cold showers. I moved to the bay area in August after nearly a 14 year stint in India. This winter the flu virus got me. I was able to fight it off without a lot of meds, I got sick after nearly 4 years.

So am not sure if my body has good immunity anymore. Overall, I have come to liking cold showers simply because of the post shower feeling.

A hot shower is great while in the shower, not so great afterwards. A cold shower is miserable the first 30 seconds, a pretty good feeling afterwards.

That's what I am gathering as well. Where is OpenAI going to have substantial proof to claim that their outputs were used ?

The reasoning prompts and answers for SFT from V3 you mean ? No idea. For that matter you have no idea where OpenAI got this data from either. If they open this can of worms, their can of worms will be opened as well.

From the R1 paper

In this study, we demonstrate that reasoning capabilities can be significantly improved through large-scale reinforcement learning (RL), even without using supervised fine-tuning (SFT) as a cold start. Furthermore, performance can be further enhanced with the inclusion of a small amount of cold-start data

Is this cold start data what OpenAI is claiming their output ? If so what's the big deal ?

Essentially software is going to become like email. Dirt cheap to produce. The folks who will get employed and continue to make money like crazy will be the email automation company equivalents. Perhaps a bad analogy but the jobs are not gonna go away anywhere.

Engineers will still continue to work like crazy and produce 100x the output. The pay could still remain the same because the profit margins on these newly developed software is gonna be so much better.

That being said, I think there will be a cycle of adjustment - may be 2-5 years for this reality to set in. So in that interim there may be joblosses.

It's not that anyone suspects OpenAI doing dodgy shit. Data flowing out of an enterprise is very high risk. No matter what security safeguards you employ. So they want everything inside their cloud perimeter and on servers they can control.

IMO no big enterprise will adopt chatGPT unless it's all hosted in their cloud. Open source models lend better to the enterprises in this regard.

I am trying out a handyman copilot for small repairs and these folks also have similar vibes. I think job protection is their no 1 priority. The field is rife with regulations as well. Some jobs need licensed professionals while other jobs don't. This varies state by state. It's a regulatory minefield from what I have seen. Perhaps rightfully so because it's your home and if something goes wrong, a lot is at stake.

It is almost always impossible to get someone to repair right away. The supply is nowhere near demand, so it is a problem worth solving IMO.

I have seen hallucinations in comments more than in code. In some of the code hallucinations, I can correct them myself. The hallucinations are obvious. try without finally blocks etc.

So my workflow is to just review every bit of code the assistant generates and sometimes I ask the assistant (I'm using Cody) to revisit a particular portion of the code. It usually corrects and spits out a new variant.

My experience has been nothing short of spectacular in using assistants for hobby projects, sometimes even for checking design patterns. I can usually submit a piece of code and ask if the code follows a good pattern under the <given> constraints. I usually get a good recommendation that clearly points out the pros and cons of the said pattern.

asymmetry is fine but not able to do a left calf raise is a problem. I do drive an automatic yes.

I have a unique related problem. My left leg calf muscle is almost an inch smaller than my right calf muscle. I just noticed it this year. Felt so weird. Why would there be imbalance in calf muscle growth to such an extreme degree. Then I tried to do a single leg calf raise on the left leg and couldn't do it. I still can't do it after 6-7 months. I must admit I am not training the left calf muscle at all, just the usual running and some sporadic squats.

This could have something to do with motor neurons not firing or something I don't know. I went to a physical therapist who suggested some calf and toe stretches and it seemed ridiculous to me. Like why would I exercise a muscle or a bunch of bones that is not functioning as expected, wouldn't that exacerbate the problem ? He didn't give a convincing answer but this paper makes me think he was spot on.

Isolating and exercising the calf muscles could probably help fire those neurons I don't know.I should do some more study on this. It's not debilitating now but I don't know what will be the impact on my balance 20-30 years from now. I am 47.

I think handwriting is a very personal trait. Some people value good handwriting and they write neat even when they write fast. Others don't. Sadly, a vast majority if the world exists in the other camp. This is why they invented typewriters. If everyone conformed to good handwriting and could agree upon a good writing style standard, the world would have been very different.

Not to mention software like peoplesoft, SAP and servicenow. Absolute shit interfaces that employees have to deal with day in and day out.

Now, an in-house tool built on top of this Anthropic API can save hours of drudgery. I can already see sales teams smiling at the new 'submit your expense report' button.

By the barrage of exits following Mira's resignation, it does look like Sam fired her, the team got the wind of this and are now quitting in droves. This is the thing about lying and being polite. You can't hide the truth for long.

Mira's latest one liner tweet 'OpenAI is nothing without it's people" speaks volumes.

I doubt someone with Mira or Ilya’s calibre have to worry about future opportunities. They can very well craft their own opportunities.

Saying I was wrong should not be this complicated, or saying we failed.

I do however agree that there is nothing to be gained and everything to be risked. So why do it.

Why is it so hard to just accept this and be transparent about motives ? It's fair to say 'we were not aligned with Sam, we tried an ouster, didn't pan out so the best thing for us to do is to leave and let Sam pursue his path", which the entirely company has vouched for.

Instead, you get to see grey area after grey area.

Predictability is great for everyone, from managers to directors to even company investors. What everyone wants to see is consistent yoy/qoq growth. We should all strive for this kind of consistency.

However, balancing this consistency with new adventures is where things get a bit tricky. I kind of like Jeremy Howard's approach to learning here. I don't know if it can be applied to a company or a team scale.

Spend 50% of your time on predictable tasks, those that you have mastery over and can do comfortably. Spend 50% of your time on frontier stuff things that break your comfort zone.

Over time the some of the latter tasks will get into the former category, thereby leading to automation organically.

The ratio of comfort:frontier tasks is personal and let's engineers choose their ratio. Some may want the ratio tilted in the comfort zone (greater predictability ) while some may choose to adventure into the frontier zone (lesser predictability ). An organization should have space for both. Even the same individual can alter these ratios based on their life stages, external circumstances.

Put rather simply, a good software engineer can also choose to be a good technician and vice versa. Why should these roles be mutually exclusive ?

Their goal might not even be to build a high growth high revenue business. Sustainable yes. So perhaps low growth, low - med revenue and who knows potentially high revenue.

If anyone else had done this, I'd be very skeptical but give Andrej's penchant for teaching and how good he is really at it, I am super excited about this endeavour.

Let's not forget the inbound and organic interest this would drive. For the intro LLM Course, people are already submitting PRs and Andrej is requesting them not to. Imagine that!

https://github.com/karpathy/LLM101n

Update June 25. To clarify, the course will take some time to build. There is no specific timeline. Thank you for your interest but please do not submit Issues/PRs.

It’s a huge problem in India. 10 times worse than US.

It's high time someone disrupted the damn desk phone network of these hospitals. It's definitely not a technical hurdle in 2024. All calls go on the data network. You route your calls out of the main router and any call that gets routed in such manner will have the ID of the router. Tag the router id to the hospital or hotel and be done with.

Is it not this simple ? With dual SIMs any phone can serve 2 lines so employees officially switch to the hospital e-sim within the hospital premises.

You should try fast.ai practical deep learning for coders part 1 and 2. It's quite dated 2022 but the principles you learn are very valid and highly useful in today's context. Especially self attention, transformers and the newer architectures based on these concepts.

Many who have done the fast.ai course have pivoted their careers into not only ML engineers but also research scientists.

It's not an easy course so to speak so you have to work through it in your spare time.

Since you are interested in deploying/scaling feel free to jump straight ahead to lesson 2 of part 1. Jeremy is an awesome teacher. I don't like or come from academia so I find his style of teaching very wholesome.

https://course.fast.ai/Lessons/lesson2.html

DJI's response (a bit dated)

https://dronedj.com/2024/03/02/dji-response-drone-ban-us/

The allegations are so subjective that they sound like some middle schooler complaining to their mom.

“DJI drones are collecting vast amounts of sensitive data – everything from high-resolution images of critical U.S. infrastructure to facial recognition technology and remote sensors that can measure an individual’s body temperature and heart rate.”

DJI's response Technically, DJI suggests using drones for body temperature checks is unfeasible.

https://www.thedronegirl.com/2020/05/06/dji-coronavirus-dron...

US politicians have totally lost their minds to even propose something like this.

Also what does "oppty" from the title mean?

The lazy man's version of opportunity

Every startup CEO must demystify 83(b) for their employees.

If you don't have cash on hand to pay for early taxes, the company can pay a signing bonus or something for those who elect 83(b) to pay for the upfront taxes.

OR

Just pay market salaries and leave the choice to employees to do whatever they want with cash. You want to buy our company stock great here's the grant. You want to put your money in S&P index go ahead.

The employee equity part needs a lot more simplification. I don't know why it is not as simple as

Here are 2 options for you

Salary 200K OR Salary 100K Equity 100K If equity 100K exercise 83(b) - pay taxes at 200K income OR defer taxes for the subsequent exercise dates. (Could land a huge tax bill) OR defer the exercise date for a liquidity event/secondary sale.

Those who value risk will take the last option and those who don't will stick to full salary. 83(b) exercise, when presented like this, doesn't seem all that rosy.

There could be some legalities that I am unaware of, but broadly this should work.

One really powerful use case they demoed was that of meeting conflicts.

"Can you meet tonight at 7?" Me "oh yes" Siri "No you can't, your daughter's recital is at 7"

It's these integrations which will make life easier for those who deal with multiple personas all through their day.

But why partner with an outside company ? Even though it's optional on the device etc, people are miffed about the partnership than being excited by all that Apple has to offer.

"The children involved in the study were born with DFNB9, which accounts for between 2 and 8% of hereditary deafness. It is caused by mutations in the OTOF gene which prevent the production of functioning otoferlin protein, which can significantly reduce sound transmission from the inner ear hair cells to hearing nerves."

Just shows how much of our body functions we take for granted. Every function has a critical path that is so vulnerable. This is despite the human body having all kinds of redundancies. Our body is so resilient, yet at times, some of the mechanisms are so fragile.

Oh yeah, you are describing a leader here. They are hustling for the next best project to take on for their team and/or for the company. They have moved beyond solving today's problems and spend most of their time on tomorrow's or the next year's problems.

What you are saying here is, for the lack of a better word, macro verification and that's completely valid.

I really don't get managers whose timetables have 100% meetings. Weekly 1:1s is at the max 10-20 hours assuming a 10:1 engineer to manager ratio.

If a manager is not technical enough to delve into an issue within their team, should they even be leading that team ? They may choose not to do so, but if a problem is not getting solved and they can't act in such a time of need, their existence is totally questionable.