ai coding -- tutor/group project/class etc..

[bruce]

hi.

meta/zuck says...

ai in meta -- killing it!!  we're having things that used to bedone by teams, being done by 1/2 people -- we're building code baby!!

ok -- so i jump off to find the great vids.. or places in the ville where this is happening!!

wheres my yoda!!

so, if you have skills in this area, and you want to share, lets see if we can talk, figure out how to build knowledge in this area.

i'm actually looking to do real world web apps, mobile apps.

and -- as an enticement, i'm willing to discuss some form of compensation..

thanks much!!

bado...@gmail.com

[Garett Gillen]

We had a subject matter expert at the December meeting. Can’t remember his name though. 

I wasn’t able to make the January meeting to see if he attended, but he may be at the February meeting. If you’re there, you might run into him.

On Jan 29, 2026, at 6:27 AM, bruce <bado...@gmail.com> wrote:



To unsubscribe from this group and stop receiving emails from it, send an email to kyoss-discus...@kyoss.dev.

[Britt Dodd]

I have some AI exposure and use it both at work and at home, it's certainly helpful in removing the monotonous tasks, allowing me to focus more on the intricate things

I use Cursor to help me fine-tune my FluxCD-on-k3s-on-KVM cluster, and i've vibe-coded some NodeJS apps to help keep my pet appointments/visits organized as well as car maintenance, it also was helpful in finishing up boring route-wiring in NodeJS for a "fake movie theater" website my 11y daughter uses when she wants to turn the living room into a movie theater.

I don't get the opportunity to go to many of the meet-ups, I think if you were to bring questions up at the meetings, Bruce, you might be able to get the answers you seek. And then you could use the money to buy us more pizza :D

-britt

[Jeff Squyres]

Here’s my $0.02 on using AI for coding: you still have to know what you are doing.

Vibe coding is great, but it only gets you so far.  It will not get you a production-grade web site or app or product.  Vibe coding can make some really impressive (small-ish) tools, or great a first take on a project.  But going the extra distance to actually make something that’s durable, reliable, scalable, … all the -able’s — that takes more than just vibe coding.

Using AI to write code can be extremely helpful (I use it all the time) when it’s augmenting what you’re doing.  I’m a grey-haired engineer; I use AI to help me do small, medium, and large projects.  I generally already know — because of experience — what the code is that I want written.  I know what I want it to do, I know what “shape” it should be, and I know how it generally should function.  Using AI makes me go faster because I already know what I want.  But it’s an important qualification: it’s always me directing the AI with my years of experience of knowing how to build something potentially complex with subtle corner cases and tricky interactions with external subsystems, rather than AI doing a task for me that I don’t understand.

There’s growing tools and methodologies that are figuring out how to use these AI-enabled tools to write large-scale applications, but with the discipline and structure that are needed for long-term projects.  Google “Spec-driven LLM development”, for example.

Relying on AI to entirely write a sizable code project without having any concept of the underlying coding / infrastructure concepts is likely to end up getting something that is flashy and/or looks good, but may not have the structural integrity that you’re going to need to actually use it and maintain it over time.  I’ve seen multiple projects that were vibe-coded by non-technical people that look great and “we just need to get into production” — but “getting it into production” basically involves a complete re-write.  I’ve seen a lot of “Vibe-coding Cleanup Specialist” labels on LinkedIn lately.

Don’t get me wrong — I’m not actually being dismissive here.  I actually endorse exactly what Britt said — his use cases sound like perfect uses of AI.  I’m just trying to frame this realistically.  Vibe coding is great as long as you recognize its limits, what it should be used for, and what it should not be used for.

-- 

{+} Jeff Squyres

[Osvaldo Rodriquez]

I have a huge doubt about how to use AI as a CS student. When I use tools like Cursor and let them write most of the code, I can build stuff really fast, but at the end, I don't know how to write that same code myself, which kills my opportunities to get a job in the field. On the other hand, when I code by myself I understand deep concepts and I feel I’m becoming a “real engineer”, but I’m slow and stay behind others who use AI, which again it moves me farther away from getting a job. So, could you guys give me some advice, my goal is to get a job but I also want to become a better engineer in the process. Thanks and I hope to see you soon at the next meeting 

[Jeff Squyres]

I agree — this is exactly the quandary for students right now.

I typically advise students to use AI for research and investigation, but:

a) do not trust it — always verify

b) do not use it to write code for you — go figure out how to do the hard thing now; go figure out how to write code now

I know there are a zillion opinions about there about how students should be educated in the age of AI.  These are just my opinions; I know others disagree with me (e.g., some advocate for just teaching students how to thoroughly use AI and don’t worry about how the code works under the covers).  I don’t personally agree, but I also thing we’re early in this are of AI — I don’t know if any of us know the right answers yet.  So having healthy debate about this with lots of different opinions is a good thing.

FWIW, here’s some slides I present in my class:

[Andrew Koroluk]

Use AI to code the stuff you already know how to do (unit tests, repetitive tasks, etc).
Don't use AI to code the stuff you want to really learn. You can use AI to explain concepts to you but don't let it write code for you if you don't truly understand what it is writing.

------

Andrew Koroluk

[bruce]

hi all!!

I want to thank everyone thats read/replied and added to this thread. I'm adding a bit of clarification. 

I've looked over sites, vids, etc.  

I could spend countless hours doing trial/error, and might get to a point where I can do something. But, I wouldn't know if it's right/better, or I'm missing something,  etc.

I wouldn't know if there's was a cheaper approach, given the input/returned tokens and the system costs, etc.

So, this is why I need to talk with someone, sit with someone while they're doing their thing to observe, ask questions!.  Anything else, I'm guessing at best.

At the same time, I'm thinking that a few others have similar needs/questions.

I'm willing to shell out some $$$ to make this happen.  

My needs are in the area of web dev. But some of the basic principles will apply to system dev, control systems, biz apps, etc...

feel free to call me

thanks

bruce

bado...@gmail.com 

512 348 9986 

[Chris Ahlstrom]

I use AI (Google) to answer relatively simple questions about code or to get a
rough idea what's up. Sometimes AI answers the wrong question. Sometimes it can
unearth wrinkles (use cases or gotchas) one has not considered.

Often I then look at the links found (Stack Overflow for example) and look at a
few solutions. This can unearth wrinkles (use cases or gotchas) one has not
considered.

Bottom line, you need ultimately to use your non-artificial intelligence to
decide what to do. In the end, I always write code that fits my "style".

-------- Andrew Koroluk 09:53 Mon 02 Feb --------

>Use AI to code the stuff you already know how to do (unit tests, repetitive
>tasks, etc).
>Don't use AI to code the stuff you want to really learn. You can use AI to
>explain concepts to you but don't let it write code for you if you don't
>truly understand what it is writing.
>
>On Mon, Feb 2, 2026 at 7:29 AM Osvaldo Rodriquez <
>osvaldo.r...@gmail.com> wrote:
>
>> I have a huge doubt about how to use AI as a CS student.
>>

>> <snip>

--
What GOOD is a CARDBOARD suitcase ANYWAY?

[Jeff Squyres]

I have no bandwidth available to do actual work in this area, but I would be willing to talk to you for 30 mins or so.  Do you have specific questions?

If so, happy to chat Friday afternoon, Feb 5; would 3pm work?

[Andrew Koroluk]

I can share some insights on AI coding agents during next Wednesday's meeting. I specifically have been using antiantigravity.googlegravity.google for a few weeks now and it is quite helpful.

[bruce]

hi Jeff.

I'm not working on an actual project.  I don't have actual specific questions.  I do have questions pertaining to the steps one uses for prompts, how prompts are implemented, how one gets better prompts vs poor prompts, which then drive costs of using the tool!

So, whenever you have any free bandwidth, I'll adapt to you.

thanks

-bruce

bado...@gmail.com 

512 348 9986