Markus Dezelak

Markus Dezelak

AI/ML Engineer • Cybersecurity Professional

Seattle, WA

Summary

AI/ML engineer and cybersecurity professional with 17+ years of operational discipline, blending deep technical expertise with modern applied machine learning. Specialized in building LLM systems, training pipelines, RAG architectures, and AI-driven security tooling. Currently completing M.S. in AI/ML while developing production-grade portfolio projects.

Technical Skills

AI/ML & LLM

Python, PyTorch, HuggingFace, OpenAI API, Claude API, LangChain, Fine-tuning (LoRA), RAG Systems

Infrastructure & Cloud

Docker, FastAPI, Redis, AWS, Hetzner, CI/CD, Vector Databases (FAISS/Chroma)

Cybersecurity

Application Security, Penetration Testing, Red Teaming, Security Automation, AI Security

Development

Go, CLI Tools, Next.js, React, REST APIs, Database Design

Education

M.S. Artificial Intelligence & Machine Learning

Colorado State University • In Progress

B.S. Cybersecurity

Bellevue University

Certificate in Workforce Education

Penn State University

Featured Projects

LLM Fine-Tuning with LoRA

Fine-tuned open-source LLM using LoRA adapters on domain-specific data with comprehensive evaluation pipeline for structure, safety, and factuality.

LoRAPyTorchHuggingFace

Production-Grade RAG System

Built retrieval-augmented generation pipeline with advanced chunking strategies, vector search optimization, intelligent routing, and evaluation framework.

RAGFAISSVector DB

AI Security & Red Teaming Toolkit

Security-focused ML toolkit for adversarial testing, prompt injection analysis, jailbreak discovery, and automated security audit workflows.

LLM SecurityRed TeamAppSec

Areas of Expertise

  • LLM solution design, fine-tuning strategies, and safety evaluation patterns
  • Cloud-hosted AI systems using Docker, FastAPI, Redis, and vector databases
  • Security-minded AI development with AppSec fundamentals and penetration testing workflows
  • RAG architecture design, retrieval optimization, and hybrid search implementations