IT/Software career thread: Invert binary trees for dollars.

Deathwing

Deathwing

<Bronze Donator>
17,853
8,793
ShakyJake said:
What is your exact workflow? Are you doing this all within a chat session?
Click to expand...
  • Pull ticket and stick it in plan mode.
  • Read plan, make corrections, ask questions.
  • Stick it in manual mode, laboriously approve all operations so it doesn't blow past code changes. Read, understand, and correct if necessary said changes.
  • If the change was substantial, spawn separate agent who coordinates a review swarm.
  • Read dozens of findings, fix some, question why others even happened in the first place. Repeat if you had to change a bunch of code, but usually jettison this loop after two iterations.
  • Create MR, review overall diff because Claude Code sucks at giving necessary context for many changes. Go back to implementation if not happy.
  • Put it up for review. Depending on who you choose: argue code comments or do another round of the previous loop(they have the same skill and just AI vomited on your MR).
  • Defang the completely separate AI review your MR receives when leaving draft mode lest the lazy human reviewer agrees with it.
I try to do most of that from Claude Code. Most of my tickets take about two thirds of Opus context by the end, not counting the review. I really dislike interacting with MRs via Claude. It's verbose to a disgusting degree and if I don't watch it like a hawk, I end up inflicting these really bloated MRs on people that don't deserve it.
 
  • 1Like
Reactions: 1 user
Noodleface

Noodleface

A Mod Real Quick
40,262
20,218
"Stick it in manual mode, laboriously approve all operations so it doesn't blow past code changes. Read, understand, and correct if necessary said changes"

This might be where we differ here. The direction we've received is tell AI to do X, let it do X and then test/review. They don't really want us caring about code changes at the basic level, but more the functional level. Obviously though, my company has different processes.
 
ShakyJake

ShakyJake

<Donor>
8,843
21,642
Deathwing said:
  • Pull ticket and stick it in plan mode.
  • Read plan, make corrections, ask questions.
  • Stick it in manual mode, laboriously approve all operations so it doesn't blow past code changes. Read, understand, and correct if necessary said changes.
  • If the change was substantial, spawn separate agent who coordinates a review swarm.
  • Read dozens of findings, fix some, question why others even happened in the first place. Repeat if you had to change a bunch of code, but usually jettison this loop after two iterations.
  • Create MR, review overall diff because Claude Code sucks at giving necessary context for many changes. Go back to implementation if not happy.
  • Put it up for review. Depending on who you choose: argue code comments or do another round of the previous loop(they have the same skill and just AI vomited on your MR).
  • Defang the completely separate AI review your MR receives when leaving draft mode lest the lazy human reviewer agrees with it.
I try to do most of that from Claude Code. Most of my tickets take about two thirds of Opus context by the end, not counting the review. I really dislike interacting with MRs via Claude. It's verbose to a disgusting degree and if I don't watch it like a hawk, I end up inflicting these really bloated MRs on people that don't deserve it.
Click to expand...
This is similar to how I work, except that I do not work inside the chat session entirely itself.

I work in several phases, and each phase produces a durable Markdown document:

- Pull the ticket or user story from ADO. This writes a `story.md`, which I then review.
- Initiate an investigation with the agent set to high reasoning. This generates `investigation.md`. I review it, resolve questions and assumptions, and agent updates `investigation.md` to reflect those decisions.
- Agent creates a `plan.md`. I read the plan and make sure the agent is not drifting off course.
- Once everything looks sound, I start implementation in steps. The agent is set to a workhorse model such as Luna 6.0 MAX. I review each step, evaluate the code, approve it, and move on until the work is done.

I do not try to one-shot implementation. The work is done in incremental steps.

I also created skills for each of these phases, and the agent follows them. I have refined the skills over time. For investigation, for example, I require the agent to write out assumptions, ambiguities, and similar issues so I can see them and make adjustments.

This has worked well for me. As I have mentioned before, I handed this workflow to a new junior dev, and she has been crushing it.
 
  • 1Like
Reactions: 1 user
Nija

Nija

<Silver Donator>
2,259
4,627
ShakyJake said:
I do not try to one-shot implementation. The work is done in incremental steps.
Click to expand...
I think this is the key.

Deathwing Deathwing I try to break the work into logical components and make sure the agent understands the phases/checkpoints I want along the way. So like the phase 1 batch of the work is performed, I inspect/iterate there and when it's done - commit. Phase 2, same thing... commit... Phase N, commit and push to create start the PR process.

I'm in alignment with ShakyJake ShakyJake in that I do an overview / planning pass (I still don't use "plan mode" as I've settled into this overview (planning I guess stage) ), and implementation "plan" phase, and finally the implementation phase, where I take it in one logical unit of work at a time. Even if that means stubbing some stuff out in the early phases before adding any meat to those methods/function calls in a later phase.

Iterate a few times in your phase 1 to make sure the document for your feature / complete unit of work is as close to perfect as you can make it. Assumptions and vague requirements here lead to a subpar implementation plan, which leads to a poor implementation. Putting a lot of thought/effort into the first few phases, in my experience at least, ends up with an implementation that is Pretty Dang Good - first shot. (first shot after a lot of iterations making sure the requirements I'm handing to the agents are crystal clear.)
 
Last edited:
  • 2Like
Reactions: 1 users
ShakyJake

ShakyJake

<Donor>
8,843
21,642
Nija said:
I think this is the key.

Deathwing Deathwing I try to break the work into logical components and make sure the agent understands the phases/checkpoints I want along the way. So like the phase 1 batch of the work is performed, I inspect/iterate there and when it's done - commit. Phase 2, same thing... commit... Phase N, commit and push to create start the PR process.

I'm in alignment with ShakyJake ShakyJake in that I do an overview / planning pass (I still don't use "plan mode" as I've settled into this overview (planning I guess stage) ), and implementation "plan" phase, and finally the implementation phase, where I take it in one logical unit of work at a time. Even if that means stubbing some stuff out in the early phases before adding any meat to those methods/function calls in a later phase.

Iterate a few times in your phase 1 to make sure the document for your feature / complete unit of work is as close to perfect as you can make it. Assumptions and vague requirements here lead to a subpar implementation plan, which leads to a poor implementation. Putting a lot of thought/effort into the first few phases, in my experience at least, ends up with an implementation that is Pretty Dang Good - first shot. (first shot after a lot of iterations making sure the requirements I'm handing to the agents are crystal clear.)
Click to expand...
Same here. I don't use "Plan" mode or really any mode other than Full Access / YOLO.

The plan is the Markdown document generated from the investigation. The point is to have a "hard copy" that the agent can follow rather than relying on its memory of the earlier conversation.

Keep in mind that context gets compacted along the way, so information can eventually be lost.
 
Deathwing

Deathwing

<Bronze Donator>
17,853
8,793
Is there that could help guide someone on this the first few times? This sounds similar to stuff Matt Pocock cooks up, but he iterates far too much to follow closely.
 
ShakyJake

ShakyJake

<Donor>
8,843
21,642
Deathwing said:
Is there that could help guide someone on this the first few times? This sounds similar to stuff Matt Pocock cooks up, but he iterates far too much to follow closely.
Click to expand...
That's kinda the thing, dude -- this is all changing so rapidly that we're all still trying to figure out how to use it efficiently. The only real recommendation I can make is to immerse yourself in it and practice using it. What we're doing today could look radically different a year from now.
 
  • 1Like
Reactions: 1 user
Nija

Nija

<Silver Donator>
2,259
4,627
Deathwing said:
Is there that could help guide someone on this the first few times? This sounds similar to stuff Matt Pocock cooks up, but he iterates far too much to follow closely.
Click to expand...
Here is what I found that kind of put the pattern I was (mostly) using to a term. This is dated, as mentioned everything is moving so fast. It's only a 15 minute video, there is a blog post somewhere that outlines things in more detail that I can't find now. But this weird lefty dude, Dexter, had (has?) some pretty good content around context engineering. Back in the OLD DAYS (less than a year ago) where the context being mostly full was a huge problem.

 
  • 1Like
Reactions: 1 user
Vinen

Vinen

God is dead
2,804
512
All I see is AI misuse causing issues in the industry right now. Seen it fuck up so much shit at customers. Kinda hillarious. I'm just waiting to get laid off so I can retire. I'd quit but 300~400K USD every 3 months is too much of a handcuff.
 
Noodleface

Noodleface

A Mod Real Quick
40,262
20,218
Vinen said:
All I see is AI misuse causing issues in the industry right now. Seen it fuck up so much shit at customers. Kinda hillarious. I'm just waiting to get laid off so I can retire. I'd quit but 300~400K USD every 3 months is too much of a handcuff.
Click to expand...
How in the hell can VMware be this profitable. We basically deleted them from our shit
 
  • 1Like
Reactions: 1 user
alavaz

alavaz

Trakanon Raider
2,032
742
Noodleface said:
How in the hell can VMware be this profitable. We basically deleted them from our shit
Click to expand...
I am also genuinely curious about this one.

We had 100s of millions invested in VMware and migrated all of it to Nutanix. Not even because the VMware quote was unreasonable... it was because we got straight up ghosted.
 
  • 1Like
Reactions: 1 user
TomServo

TomServo

<Bronze Donator>
9,021
21,518
alavaz said:
I am also genuinely curious about this one.

We had 100s of millions invested in VMware and migrated all of it to Nutanix. Not even because the VMware quote was unreasonable... it was because we got straight up ghosted.
Click to expand...
Same. We dumped their dogshit
 
  • 1Like
Reactions: 1 user
TJT

TJT

Mr. Poopybutthole
<Gold Donor>
47,906
131,971
Noodleface said:
"Stick it in manual mode, laboriously approve all operations so it doesn't blow past code changes. Read, understand, and correct if necessary said changes"

This might be where we differ here. The direction we've received is tell AI to do X, let it do X and then test/review. They don't really want us caring about code changes at the basic level, but more the functional level. Obviously though, my company has different processes.
Click to expand...
I am currently doing a big migration. As I said I work in data infrastructure. One pillar of that at many companies is Airflow. I have been asked to lead the project migrating from V2 to V3 across the data platform. Which I have done several times before.

This company has >50 Airflow instances and thousands of DAGs (orchestrated python script that does some data operation). Doing such a migration would take years if we started it 5 years ago. Today?

I fire up a Cursor Project or Plan, tell if I need to migrate every single DAG in this project to a V3 format. I also have a list of best practices in terms of design that I want it to do. When it changes like 300 of the things I do have to go through it and examine a lot of them. I of course have to test and verify it actually functions in the instance for whatever reason. But this whole project will now take me probably 3 months. Working on it mostly alone. This place uses Astronomer which I have not really used in an enterprise capacity before but god damn. If you use Airflow and don't use Astro you are a fucking retard unless you just have one single small instance of it. I am also able to get all of this shit onto Azure from AWS in one fell swoop here too.

Even the most superhuman 1000x developer could not even scratch the surface of such a project on his own in a few months before AI. Not even one of these Airflow instances.

It is absolutely wild man.
 
Last edited:
  • 1Like
Reactions: 1 user
Noodleface

Noodleface

A Mod Real Quick
40,262
20,218
TJT said:
I am currently doing a big migration. As I said I work in data infrastructure. One pillar of that at many companies is Airflow. I have been asked to lead the project migrating from V2 to V3 across the data platform. Which I have done several times before.

This company has >50 Airflow instances and thousands of DAGs (orchestrated python script that does some data operation). Doing such a migration would take years if we started it 5 years ago. Today?

I fire up a Cursor Project or Plan, tell if I need to migrate every single DAG in this project to a V3 format. I also have a list of best practices in terms of design that I want it to do. When it changes like 300 of the things I do have to go through it and examine a lot of them. I of course have to test and verify it actually functions in the instance for whatever reason. But this whole project will now take me probably 3 months. Working on it mostly alone. This place uses Astronomer which I have not really used in an enterprise capacity before but god damn. If you use Airflow and don't use Astro you are a fucking retard unless you just have one single small instance of it. I am also able to get all of this shit onto Azure from AWS in one fell swoop here too.

Even the most superhuman 1000x developer could not even scratch the surface of such a project on his own in a few months. Not even one of these Airflow instances.

It is absolutely wild man.
Click to expand...
Yeah man it's nuts. I have a very stubborn senior dev that's not really using AI (he's autistic and he thinks he's right about EVERYTHING) who argues with me about improvements or optimizations I want to make. So he went on vacation this week and I took it upon myself to plan a full refactor. Created about 45 stories with hard evidence, what code to change, and what exactly the impact is. Plenty of things he's said over the years "it's impossible" either because he investigated or he just thinks it.

This is work I would NEVER take on in the middle of a server generation, huge risk and a lot of time to sink into this. Claude has turned it into something we can spin in a few weeks n
 
  • 1Like
Reactions: 1 user
Cad

Sheriff Cad

scientia potentia est
<Nazi Janitors>
35,470
85,799
Can I just say the difference in the way you guys describe how you use AI (and the results) and the stuff posted in the AI threads is like night and day difference.
 
You must log in or register to reply here.