Mykyta Mudryi

Co-Founder @ ARIMLABS

Warsaw, Poland, CET

XLinkedIn
MM

About

Co-Founder at ARIMLABS, driving the strategic vision for AI security research and product development. Expert in discovering critical vulnerabilities in AI systems, developing secure-by-design methodologies for AI agents, and scaling cutting-edge R&D operations. Proven leader with deep technical expertise across cybersecurity domains, including hands-on experience auditing Fortune 500 infrastructures and contributing to OSS-Fuzz.

Certifications

BSCP ↗️
eMAPT ↗️
eWPTXv2 ↗️
NVIDIA: EAML ↗️

Skills & Expertise

Security Research

Vulnerability Research
Binary Fuzzing
Source Code Security Review
Penetration Testing

AI/ML Security

LLMs Security Assessment
Adversarial Machine Learning
AI/LLM Safety
AI Agent Security

Programming

Python
C/C++
Rust

Leadership

R&D Management
Security Leadership
Strategic Planning

Work Experience

ARIMLABS.AI
Remote

2024 - now

Co-Founder

ARIMLABS is a private security research lab driven by a small but highly talented team dedicated to making the AI world safer than it was yesterday. Our mission is to create a secure and confidential AI ecosystem for everyone through rigorous security audits and innovative internal software solutions.

Leviathan Security Group | Europe
Full-time

2022 - Now

Associate Security Consultant → Middle Security Consultant

Highly skilled in identifying security issues in complex projects, with a proven track record of effectively managing team operations during project execution.
Led an R&D project on binary fuzzing, developing an internal tool that utilizes a custom-trained LLM and mathematical models to reduce fuzz target identification & development time from 3 days to 45 minutes.

Education

Polish-Japanese Academy of Information Technology | BSc

2022 - 2026
  • Bachelor's Degree in Computer Science, awarded Rector's Scholarship for academic excellence
  • Bachelor thesis topic: An application of Graph Neural Networks in static source code analysis
  • Works under the supervision of prof. dr hab. Grzegorz Wojcik

Featured Publications

The Hidden Dangers of Browsing AI Agents

2025
M. Mudryi, M. Chaklosh, G. Wójcik
This paper presents a comprehensive security evaluation of such agents, focusing on systemic vulnerabilities across multiple architectural layers. Our work outlines the first end-to-end threat model for browsing agents and provides actionable guidance for securing their deployment in real-world environments.
AI Agents
Adversarial AI
CVE

Languages

Ukrainian (native)
English (C1)
Polish (B1)
Deutch (A1)

Open Source Security Contributions

Apple (MLX) ↗️

Under my guidance, ARIMLABS’ autonomous security pipeline uncovered two previously unknown memory-safety vulnerabilities in Apple’s open-source MLX framework — CVE-2025-62608 and CVE-2025-62609.

AI Agents
Vulnerability Research
CVE

Browser Use ↗️

CVE-2025-47241

Vulnerability Research
CVE

Libzip ↗️

During the contribution, 4 fuzz targets were developed, leading to the discovery of several security issues.

Binary Fuzzing
C/C++
Source Code Review
OSS-Fuzz Google

Pandas ↗️

During the pull request to the OSS-Fuzz repository, seven fuzz targets were developed, which led to the disclosure of an unhandled exception.

CPython
Binary Fuzzing
Security Source Code Review