Varna, Varna, Bulgaria
I'm a Cybersecurity researcher with a Master's degree, focused on AI Red Teaming and LLM Safety. I built CL-RAM (Cross-Lingual Research Assessment Module), a Python framework that automates adversarial testing of Large Language Models using MITRE ATT&CK tactics. It runs 10 LLM validators in consensus to evaluate attack success rates. My thesis covered 1,680 automated test cycles across Mistral-7B, EuroLLM-22B, and Phi-4. I also develop iOS apps. My Personal Habit Tracker is currently in App Store Review. I've built a Universal ECG Viewer with CUDA GPU acceleration and a ResMed CPAP therapy data viewer. Before cybersecurity, I studied Maritime Navigation and Astronomy. I participated in four IASC asteroid search campaigns using Astrometrica software and discovered multiple asteroids designated with my initials (AZD). I also published a research paper at IAMU AGA 18. Right now I'm exploring PhD opportunities in Cybersecurity, building on my CL-RAM research. What I work with: AI Security: Red Teaming, Prompt Injection, Model Alignment, RLHF, Jailbreaking Engineering: Python, PyTorch, CUDA/CuPy, SciPy, NumPy, Dart/Flutter, Transformers Mobile: Flutter, StoreKit, CloudKit, iOS App Store Connect Blockchain: Ethereum, Solidity, Rust, Solana Open to opportunities in AI Safety, Red Teaming, and Security Engineering.
Project: CL-RAM Developing an automated framework for adversarial testing and red-teaming of Frontier LLMs. Automated Attack Engine: Engineered an end-to-end Python framework executing MITRE ATT&CK vectors to test model robustness against jailbreaks and prompt injection. Tooling & Infrastructure: Designed and built "Thesis Validator Pro" – a custom desktop GUI application (Tkinter/SQLite) for high-precision human-in-the-loop verification of 8,000+ attack vectors. Advanced Validation: Implemented a consensus-based validation engine orchestrating 10+ calibrated LLMs to reduce false negatives in safety benchmarks. Tech Stack: Python, PyTorch, llama.cpp, Pandas, SQLite, Git.
Managed complex operational logistics, safety compliance, and critical incident response in a high-stakes industrial environment. Crisis Management: Led decision-making during time-sensitive operational incidents, ensuring personnel safety and strict adherence to international protocols. Process Optimization: Analyzed workflow inefficiencies and implemented strategic improvements to optimize turnaround times and resource allocation. Safety & Compliance: Enforced rigorous safety standards and risk management protocols—competencies directly transferable to AI Safety and Policy enforcement. Leadership: Coordinated cross-functional teams and communicated operational requirements to international stakeholders.