iguana

Brilliant idea and great execution!

Trivial examples show that this isn't nearly as good as ChatGPT. The headline should be changed.

What makes you think copy and pasting responses from an LLM is the actual job being posted?

This is demonstrably false for many use cases. For one broad example, LLMs have shown incredible performance on many NLU and NLP tasks, that are not currently possible using other techniques.

That's why the word "engineer" is in the title and it isn't "prompt writer" or "ChatGPT user".

It's highly unlikely that anyone taking this effort seriously is copying and pasting from ChatGPT, rather than using the API and building pipelines as part of a broader system.

Instruction-fine-tuned LLMs like ChatGPT require creating, validating, and maintaining prompts. Finding ways to use them safely is also not easy - prompt injection and hallucination are just 2 potential pitfalls - there are many more.

Denigrating this effort as "AI monkey" is myopic at best, but really just comes across as a signal that someone is terrified of being replaced by this new tech. With that attitude, they will be.

Writing prompts and engineering together = prompt engineer. The engineering depends on the prompt and the prompt depends on the engineering. Just like an ML engineer, or a QA engineer, or [anything] engineer. How specific the job title gets really depends on hiring criteria and daily job function.

Otherwise, the job title would be "prompt writer".

Your point is what, that existing engineering titles cover this effort? Sure, you can just call all of it software engineering, but sometimes it's useful to be more specific. The LLMs are so powerful now that this new, more specific title makes sense to me, and clearly those using this new title. We'll see how it pans out over the next few years.

This is an unhelpfully cynical take. The job title has "engineer" in it, so a more charitable interpretation is more serious than "AI monkey".

Using LLMs to solve real problems is not easy. Making sure that you don't introduce regressions while making improvements is difficult, and requires building and evaluating a dataset, and the necessary pipelines. It may also include diversification of LLM providers, and creating the necessary abstractions. A fundamental understanding of how LLMs work, ability to compare different architectural approaches, along with typical data engineering and software development skills would be required.

What if you want to use the LLM for Question/Answer systems that requires working with embeddings? What if you want to find a way to process data locally without sending sensitive data to the LLM provider?

This requires real engineering skills.

For those with the new Samsung Galaxy S23 series, they use a Snapdragon X70 modem and so are not impacted.

If you're interested in increasing EVOO consumption for the health benefits based on polyphenols, look into Moroccan olive oil, as it has very high polyphenol levels - as high as 30x regular EVOO.

You may be conflating programs running in a terminal and the terminal itself. We've managed to get this far without the latter.

Replicant | https://replicant.ai/ | QA Lead, Fullstack, Deep Learning, Data Engineering, and Telephony Engineering positions | San Francisco, CA or REMOTE Replicant is a Conversational AI technology that works out of the box to solve customer problems over the phone. We craft great conversations by combining Machine Learning, Artificial Intelligence, and linguistic conversational design into the fastest, smartest, and most expressive Thinking Machines you’ve ever spoken with.

We're a small team (~20 people) tackling a big industry with many eyes on it, using powerful technology. Our team comes from a diverse background of industry and the arts, and we are distributed across the US and Canada.

Our stack includes TypeScript (browser and node), JavaScript, Postgres, Redis, Python (3.x), and Pytorch Infrastructure is Google Cloud (though we also use AWS and Azure) and most services run in k8s (Kubernetes)

We're hiring:

* QA Lead - own quality engineering, E2E testing, automation, and testing conversations on the phone

* Deep Learning / NLP / Transcription - Transformers, Intent detection

* Data Engineering - model data and build data pipelines for realtime low latency inference

* Telephony / DSP Engineer - SIP integrations, low latency audio processing

In order to support you we offer:

* A remote-friendly culture: Communication is big. Most of us work remotely, full and/or part time.

* Offsites: We come together regularly for some unwinding and face-to-face time.

* Benefits: a great health plan, equity, and 401K.

However, the most significant advantage is that you'll be early enough to shape Replicant's culture and the next era of growth.

Please reach out to: jobs@replicant.ai

Replicant | https://replicant.ai/ | QA Lead, Fullstack, Deep Learning, Data Engineering, and Telephony Engineering positions | San Francisco, CA or REMOTE

Replicant is a Conversational AI technology that works out of the box to solve customer problems over the phone. We craft great conversations by combining Machine Learning, Artificial Intelligence, and linguistic conversational design into the fastest, smartest, and most expressive Thinking Machines you’ve ever spoken with.

We're a small team (~20 people) tackling a big industry with many eyes on it, using powerful technology. Our team comes from a diverse background of industry and the arts, and we are distributed across the US and Canada.

Our stack includes TypeScript (browser and node), JavaScript, Postgres, Redis, Python (3.x), and Pytorch Infrastructure is Google Cloud (though we also use AWS and Azure) and most services run in k8s (Kubernetes)

We're hiring:

* QA Lead - own quality engineering, E2E testing, automation, and testing conversations on the phone

* Deep Learning / NLP / Transcription - Transformers, Intent detection

* Data Engineering - model data and build data pipelines for realtime low latency inference

* Telephony / DSP Engineer - SIP integrations, low latency audio processing

In order to support you we offer:

* A remote-friendly culture: Communication is big. Most of us work remotely, full and/or part time.

* Offsites: We come together regularly for some unwinding and face-to-face time.

* Benefits: a great health plan, equity, and 401K.

However, the most significant advantage is that you'll be early enough to shape Replicant's culture and the next era of growth.

Please reach out to: jobs@replicant.ai

How confident are you that the results will indicate a level of impairment, rather than a particular concentration of THC?

Other substances contained in cannabis have significant synergistic effects with THC; are you going to look for other cannabinoids and the presence of terpenes?

I don't have a book recommendation for you (what helped me was the standard documentation and about a year of daily dev time), but an observation:

Your team is likely using an overly restrictive tsconfig.json. No-implicit-any, for example.

Make it less restrictive, and you then have typing and no penalty for punting on it until later.

If using an IDE, the effort into defining types is immediately rewarded with autocomplete for API/backend AND client code, and we even export the types for use by a Monaco editor inside an internal app.

This is great!

One way to offset the cost for job seekers is to take some part up front, and the rest in ~90 days when they get hired.

I'm in the same boat. At this point, "growing pains" is not a legitimate excuse for this level of incompetence. I (wrongly) assumed that since Coinbase is based in SF and funded by YC, that they would not be able to get away with this level of negligence. This reflects very poorly on YC and specifically YC leadership.

Serious question for YC insiders: has YC leadership made any effort to advise Brian on this, and how poorly this negligence reflects on YC as a whole, or does no one care because they're getting such a high return on their investment?

Hello Mr. Schneiderman,

I hope that by "AMA" you really do mean, anything :)

What is your take on the outdated gravity knife law?

Lots of advice here is to go after pain points, or enter competitive markets, or to question your team dynamics. I have a different perspective:

Set up a studio, where the team builds a new "thing" every 2 weeks. Then, 2 months later you have 4 things, and you can pick the one that inspired the team the most.

Given your cash position, you should be able to easily float 2 months while building a bunch of different things.

Sure, I'd use it, but under the following conditions:

- wired ethernet

- lots of guard code that reconnects failed calls

- customer is aware they will not get 5 9's or even 3 9's.

Typically using WebRTC means a significant cost savings, so it may or may not be worthwhile to build out the additional stuff needed to make it work.

I'm basically a twilio engineer now. I do a lot of telephony, but no one wants to run and scale their own stack, so everyone ends up building against REST apis. Twilio has some of the best features around reliably placing calls, sending texts, and the most enjoyable developer experience of any api I've ever used, including outside of telephony. I've built a 100% serverless contact center, among other things. Twilio Sync RT infra is also fantastic. Their WebRTC service was total shit through 2014 but is better now.

Thank you for replying!

The sub-mm scale would be to accurately model small things. When making a case for a raspi0, for example.

First, thank you! I use the ipad version.

Some needed features:

- precise editing

- manual entry of dimensions

- different materials and transparency

- library of common items such as batteries, usb ports, etc

- sub-mm scale

- pencil support for ipad pro

Operationally, the CEO should probably not have write access to a master DB.

This is really cool!

One issue I found is that it provides alternative spellings as distinct items. Is there a workaround for this?

{ "typeOf": [ "chromatic color", "chromatic colour", "spectral colour", "spectral color", "citrus", "citrus tree", "pigment", "citrous fruit", "citrus fruit" ] }

This is the best headline I've ever read.

This is long term/platform play. Think wordpress. The problems they are solving are not easy to solve. They are executing well! Worst case, one of snoozefest enterprise shops will just buy the whole thing.

Would it be possible to import into the US?

This is going to be awesome for cheap "phone home" drone projects. I wonder if the sims would work in a mobile hotspot (and just allow traffic to *.twitter.com) or some other programmatic access. I would pay $12/month for my dronrs to tweet.

It could be used as part of the process, for example to access a URL with an android exploit on it that the user would not normally visit. It could also detect periods of inactivity, or high background noise. This is a real attack vector not worth dismissing.

This is completely ridiculous and written by someone with absolutely no experience with marijuana. The idea that you "could end up high years later" is laughable, and has absolutely no basis in scientific fact.

"A stoned Doctor might not smell of it, might not have any real outwards signs, and then could get the giggles when he nicked an artery."

Marijuana does not cause a significant impairment in regular users. Even if that were not the case, if a doctor chooses to be intoxicated when they are responsible for someone else's life, they are already making a terrible choice, regardless of the legality of the substance in question.