lfciv

What is this?? Permutation City?!

If you do end up getting a used Apple device I recommend making sure that it has the newer keyboard style that the current models have (I think it switched in 2018). The old keyboard was not fun to use.

Working with various data, ML, and researcher-type engineers there's a bit benefit to using Python. Especially if you want to be at the forefront of what's being built.

I've been building with Python FastAPI and a React/Nextjs app. I've gone back and forth on whether we should have used Django, but we've been full steam ahead with what we have.

If I wasn't building in the AI space I think I probably would have used Node <-> React with tRPC to make the typed full stack monorepo experience seamless.

Oh and Postgres - I'm irrationally loyal to Postgres

I feel like work from home is the modern version of this. If I need to go heads down I'll just let the team know and work from home.

Based on nothing but a guess, my bet is that it's probably roughly breakeven

Though I'm curious how they're not charging any subscription given that the device will have ongoing costs to run the models. That feels like it could hemorrhage money – though if users are actively using it I guess it'd be a good problem to have

Given your timeline (since the 1940s) I'd say cell phones in general come to mind as meeting the hype

I also think Cloud Computing has generally met the hype

A bit of a different take, but given the amount of experience you have, I wouldn't put as much weight on using a CV at all.

With an updated LinkedIn you can reach out to past bosses, coworkers, and friends for introductions to other people in industry (whether they're hiring or not) – and follow that chain until you find something you're interested in.

It feels like it's sort of it's own thing. LLMs are really good at morphing or fuzzy finding.

An interesting example – I had a project where I needed to parse out addresses and dates in a document. However, the address and date formats were not standardized across documents. Utilizing LLMs was way easier then trying to regex or pattern match across the text.

But if you're trying to take a text document and break it down into some sort of a structured output, the outcome using LLMs will be much more variable.

Companies with an API-first product will have a lot of opportunities to do non-web work.

Bit of a tangent, but I always come back to this read:

https://www.stilldrinking.org/programming-sucks

Can you elaborate on what you mean by "really complex solution"? It is well-written/designed, but over-engineered? Or is it poorly done and subsequently hard to read + maintain? Or is it moreso an issue of documentation?

Also regarding CRUD apps – I actually was able to automate code generation of most of my CRUD pretty easily using cursor. I hand wrote a couple of them and now I just create the new file and prompt it referencing the Model and one of the other CRUD files as an example.

I used to use Github Copilot, but I find Cursor (https://cursor.sh/) to be even better.

You can add an OpenAI API Key and it'll use the GPT-4 API. It indexes your code for you and it'll crawl documentation for you if you give it the link. Then you can reference specific files or documentation with "@". There are also some really intuitive hot keys.

> I agree, but are HR people agreeing on this?

Really depends on the company I guess

> Is it sufficient to be skilled in one programming language, such as Python, and implementing a software renderer or ray tracer, your own neural network, your own HTTPS server, your own OS, your own memory allocator, your own interpreter, your own virtual machine...?

May be misunderstanding here, but if I was implementing my own OS I probably wouldn't be using Python. Different problems warrant different solutions – by a "general problem solver" I mean that you think about the coding language as a tool to solve a problem. Similar to how you might think of S3 or ECS as a tool.

That you're thinking of the problem in the business context rather than purely the technical context.

Full stack engineer – and doing more AI engineering these days

GPT-4 has fully altered my dev workflow. I don't think I could move away from it unless there was a clear, equivalent, alternative.

Not only on the OpenAI UI, but also via the Cursor IDE which I actively use with an OpenAI API Key.

Even if you don't use the OpenAI UI you'll likely end up using the produce via other products. Github Copilot, Arc Browser, Duolingo – likely more in the future – all use GPT under the hood.

Not exactly in a DAW, but I found Syntorial to be incredibly helpful. Teaches you a lot of jargon around sound manipulation and how to reproduce a desired sound. It'll make you way more comfortable navigating a DAW later.

https://www.syntorial.com/

I view it similarly, but differently. I think at some point a developer should aspire to level up to being a general problem solver.

The language, the stack, etc.. are just details. Given a problem - they can tackle it.

I've been thinking about this recently regarding photos – Would be sad to lose a life's worth of photos just because someone can't get into your phone.

In a similar, but different vein, I've recently found interesting to think about how all the photos of yourself tend to live on other people's phones – whereas photos on your own tend to be of places and other people.

Back when they used to charge a few dollars to download the app I bought the Ultimate Guitar App (https://www.ultimate-guitar.com/)

Honestly been pleasantly surprised now that they've had a subscription model for years and have honored my app purchase from probably over ten years ago

Even if they're non-technical I value a technical approach to problem solving.

I've always liked the idea that if they didn't have a technical cofounder they might still be able to hack together an MVP using tools like Bubble, Zapier, Airbase, Sheets, etc.

Always has been...

I'd be curious if there are any strong programs where devs can provide mentorship to students interested in potentially becoming developers down the road (especially underrepresented). I find when you're just getting started there's a confidence issue where having a mentor rooting you on makes a huge difference.

One thing I love about the company I'm currently at is that pretty much every eng manager does some level of IC work. Meaning they all have some pulse on the codebase beyond higher level planning.

Also in a weird way my BeReal has become a bit of a photo diary – I use it pretty consistently on a daily basis. It's crazy how much looking back at a single photo can jog your memory on what you did that day.

Since reading "The Making of Prince of Persia" (highly recommend) I've tried to journal – roughly once a day – just a way to recount my day to day in the future when I don't remember the smaller details.

I've been recently thinking of this as similar to Data Warehouses vs a normal Relational Database. The Data Warehouse has a variety of usecases around analytics or marketing where it has replaced RDBs, but for other usecases – especially around your core application transactions – you still need your Database.

Similarly, if you're doing something like generating medical/therapy notes, or generating SAT questions then GPT is great, but for transactions or cases where exact behavior is a requirement, then you sort of need a discrete set of instructions that only a programming language can provide. You're not going to see payments systems built by GPT in the short term at least.

I wonder if any earlier prompts asked about Honda – feels a bit weird to ask what the cheapest car overall is and have the chat bot say "well let me tell you about some Hondas overall"

And if the user did specify Hondas then why would Honda pay for it? Wouldn't the Chat bot give the same answer regardless of whether it's paid? I guess you pay for the CTA link?