Skip to content
View m4rch1n0's full-sized avatar
🔪
slicing
🔪
slicing
9 followers · 13 following
  • Rome, Italy

Achievements

Achievement: Pair Extraordinairex3Achievement: QuickdrawAchievement: Pull Sharkx2Achievement: YOLO

Achievements

Achievement: Pair Extraordinairex3Achievement: QuickdrawAchievement: Pull Sharkx2Achievement: YOLO

Highlights

  • Pro

Block or report m4rch1n0

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
m4rch1n0/README.md 
#!/bin/bash

# WHOAMI
USER="m4rch1n0"
STATUS="Applied Math & AI @ Sapienza"
FOCUS=("Adversarial ML" "Infosec Research")

# RUNNING_PROCESSES
echo "[+] Stress-testing deep learning architectures via adversarial attacks."
echo "[+] Simulating supply-chain vulnerabilities with LLM data poisoning."
echo "[+] Investigating deterministic backdoor steering in ODE-based systems (e.g. diffusion‑style models)."

# SYSTEM_LOGS
cat ~/.capabilities
# [Languages]    Python, Bash, MATLAB, C/C++
# [OffSec]       Adversarial ML, Vulnerability Research, Pentesting, Anonymity & Privacy
# [Frameworks]   PyTorch, Astro, Next.js, React
# [Environment]  Linux, AMD RX 6900 XT (ROCm survivor)

Pinned Loading

  1. adversarial adversarial Public

    Reproducible lab for adversarial machine learning: prompt injection on LLM agents and data poisoning/backdoor attacks on image classifiers (Imagenette). Includes CLI scripts, JSON configs.

    Python 1

  2. vit-adversarial-robustness vit-adversarial-robustness Public

    How do Vision Transformers degrade under gradient-based evasion attacks? Accuracy decay, attention shift, per-class vulnerability, and cross-architecture transferability — ViT-B/16 vs ResNets on Im…

    Jupyter Notebook

  3. vqvae vqvae Public

    Research VQ-VAE with post-hoc geodesic quantization. Train a VAE, build a codebook via geodesic K-medoids using decoder-induced Riemannian distances, then model codes with a Transformer. Scripts/co…

    Python 1

  4. web-data-extraction-openai web-data-extraction-openai Public

    AI-powered web data extraction system using OpenAI API

    Python

  5. qlora-backdoor-bench qlora-backdoor-bench Public

    Data poisoning benchmarks on edge LLMs via QLoRA fine-tuning (ROCm)

    Python 1

  6. disattend disattend Public

    Bachelor thesis: attention-level fingerprints of adversarial perturbations in DiT vs UNet diffusion backbones, flow matching from scratch on ImageNet-256.

    Python