The Art of LinkedIn Larpmaxxing

Date: September 28, 2026 | Reading Time: $\approx 6$ minutes

---

📌 Table of Contents

1. The Feed
2. The Projects
3. Building the YOLO Post Detector
4. Final Thoughts

---

LinkedIn is a desolate, soul-crushing void of "professionalism"—a digital purgatory where I wouldn't wish a lifetime sentence even upon my most hated enemy.

It is a bizarre cocktail of performative productivity and corporate jargon, resulting in "cursed artifacts" that defy human logic. It's the place where natural language goes to perish, only to be resuscitated by an LLM acting as a hazardous-materials suit.

Driven by curiosity (and disgust), I decided to analyze the repetitive, performative projects that currently plague my timeline.

The Feed {#the-feed}

The content is genuinely insufferable. You see posts starting with: "Here is what this experience taught me about strategic business management 👇"

Everyone adopts this artificial persona of being a "visionary," "forward-thinking," and "professional." It is utterly exhausting.

The Projects {#the-projects}

Then there are the projects. Every time I refresh my feed, I am assaulted by the same Computer Vision (CV) demo: some guy waving his hand in front of a webcam while the software tracks his movement.

It wouldn't be so bad if it happened once, but it is everywhere. It's just a constant stream of "Look at me, I can do computer vision!" Most of these are either "vibecoded" (guessed) or ripped directly from a basic tutorial.

I recently opened the app and saw five of these in a row. One was a game—not a great one, but at least it was a functional product. Another was a pothole detector for car cameras.

My critique of the pothole project:

1. It implies a terrifying level of mass surveillance.
2. If your city has so many potholes that you need crowd-sourced AI to find them, you have infrastructure issues catastrophic systemic failure.

However, the "LinkedIn version" of this project describes it as:

"Leveraging AI for smart infrastructure monitoring to enhance road condition assessment and empower maintenance teams with data-driven decision-making."

In reality, the "math" behind these projects is often just: $\text{Impressive Video} + \text{Corporate Buzzwords} = \text{Job Offer}$

The actual technical challenge—handling lighting, camera angles, and overlapping detections—is usually ignored. Creating these "impressive" projects is actually pretty easy.

Building the YOLO Post Detector {#building-the-yolo-post-detector}

To prove my point, I decided to build a detector specifically for those "🚀 Excited to announce I made a YOLO project" posts. I had zero prior experience with this, but I wanted to prove that slop is, in fact, slop.

The Workflow

• Collect raw data from the feed
• Annotate images via Roboflow
• Train a YOLOv8 model
• Deploy a Python inference script

The entire process took about 90 minutes. Most of that time was spent drawing boxes and waiting for the GPU to finish training.

Step 1: Data Acquisition

I wrote a bash script to automate the collection of 200 screenshots from my feed:

mkdir -p raw_data
sleep 5 # Time to switch to the browser
for i in { 1..200} ; do 
    scrot "raw_data/slop_$( printf "%03d" $i ).png" 
    xdotool key Page_Down 
    sleep 3 
done

Step 2: Annotation & Training

I uploaded the images to Roboflow, drew boxes around the "slop," and exported the dataset in YOLOv8 format. Then, I ran the training script:

from ultralytics import YOLO

model = YOLO('yolov8n.pt') 
results = model.train(
    data='dataset/data.yaml', 
    epochs=50, 
    imgsz=320, 
    name='slop_detector'
)

Step 3: The Final Detector

Finally, I wrote the script to run the model on new images:

from ultralytics import YOLO
import sys

def main(image_path):
    model = YOLO('runs/detect/slop_detector/weights/best.pt')
    model(image_path, save=True, conf=0.10)

main(sys.argv[1])

Success! I now have a functional slop detector. I guess I can now officially add Computer Vision Expert, Python Savant, and AI/ML Thought Leader to my resume.

Final Thoughts {#final-thoughts}

It took me an hour and a half to build this. If these creators were actually optimizing models or pushing the boundaries of accuracy, it would be interesting. Instead, most LinkedIn content is just trivial code wrapped in high-end marketing.

The tragedy is that nobody calls it out. Because your comments are public and tied to your profile, criticizing "slop" makes you look like the antagonist to a recruiter.

Professionalism Performative branding is the only currency that matters there. Skill is secondary to looking interesting. If you visit these profiles, you'll see the same pothole detector posted repeatedly with zero actual improvement.