@noob6t5

Sangharsha

@noob6t5Joined October 2024

Dev & Tester

About

I write to revisit topics I’m interested in or when I’m bored and curious.

Available for

Vulnerability Testing - Web2 ,AI/ML (my bot's and agent's) Smart Contract Auditing - Learning

Sangharsha's blogs

noob6t5noob6t5.hashnode.dev37 posts

Articles Comments

Recently published

SSangharshanoob6t5.hashnode.devJun 27 · 1 min read

`ls` That Saved Me From Deleting Weeks of Work (AGAIN)

I almost deleted 3 weeks of work coz I got a little too confident with a shell wildcard.I was cleaning up my recon workspace and wanted to remove a bunch of generated directories. rm -rf {reports,logs

SSangharshanoob6t5.hashnode.devJun 11 · 4 min read

Pickle Deserialization RCE via Model Upload Endpoint

Link: https://www.ratctf.com/challenges/synapse-lab This challenge revolves around a classic but still heavily abused primitive: unsafe Python pickle deserialization exposed through a model upload API

SSangharshanoob6t5.hashnode.devMay 12 · 3 min read

Beyond the Canvas: Exploiting an Unsanitized Side-Channel in draw.io

I don't usually write about my web findings, but I felt this one needed to be documented as a reminder for my future self too. My initial mistake was tunnel-visioning on a direct bypass of the main sa

SSangharshanoob6t5.hashnode.devApr 14 · 7 min read

How OPT(ADD).Mathematics Builds the Foundation for AI/ML World

In this article, I have tried to connect each chapter with insights from teaching and real-world experience in building and analyzing AI systems, showing how these concepts power modern technology and

SSangharshanoob6t5.hashnode.devApr 12 · 3 min read

Deceptive AI Doesn’t Break the Rules. It Optimizes Around Them.

Everyone assumes deception in AI will look obvious. A glitch. A spike. A detectable anomaly. That assumption is wrong. The most effective deception doesn’t violate the system.It stays statistically i