tinywifi

student. i build things and document what i find when i poke around.

into computer vision, security research, and understanding how systems fail. based in thailand.

security researchcomputer visionstudentthailand 🇹🇭

read the blog say hi pgp

~/bloglatest ->corn kernel grading

burp — repeater

POST /api/forgot-password HTTP/1.1
Host: [redacted].ac.th
Content-Type: application/json

{"email":"victim@school.ac.th"}

HTTP/1.1 200 OK
Content-Type: application/json

{
  "status": "ok",
  "reset_token": "a3f9c2d1..."
}

grain-clf — inference.py

# YOLOv11 corn seed grading pipeline
model = YOLO("grain_clf_v11n.pt")

# USDA damage classes
CLASSES = [
    "fungal", "insect",
    "size_irregular",
    "low_germination",
]

results = model.predict(
    frame, conf=0.65, iou=0.45
)

~ — bash

❯ whoami

tinywifi

❯ cat about.txt

student · security · computer vision

pokes at how things work

documents what breaks

❯ echo $STATUS

actively learning ▌

~ — bash

❯ whoami

tinywifi

❯ cat about.txt

student · security · computer vision

pokes at how things work

documents what breaks

❯ echo $STATUS

actively learning ▌

1/3

drag or click to cycle

tech i use

PythonC++JavaJavaScriptGLSLFabricLinuxOpenCVBurp SuiteWiresharkYOLOv11PyTorch

projects

CanvasGLSL

minecraft fabric mod that replaces menu backgrounds with custom GLSL shaders. includes an in-game shader IDE, syntax highlighting, Shadertoy-compatible uniforms, live/manual compile modes, and a file manager for organizing shaders.

helium-sync Windows

bidirectional sync of Helium browser bookmarks + saved tab groups across Windows machines, using your own private git repo as transport. CLI, no extension. Windows fork of aadarwal/helium-sync (was macOS-only).

grain-clf

real-time corn seed quality grading via a fine-tuned YOLOv11 pipeline. detects fungal contamination, insect damage, size irregularity, and germination defects from a single image in under a second. validated against USDA damage-class standards.