I'm a graduating senior at Drake University (Class of 2026) pursuing dual B.S. degrees in Computer Science and Artificial Intelligence with minors in Cybersecurity and Information Systems. Starting June 2026, I'll be joining UL Solutions as a full-time AI Incubator Engineer.

My interests span AI/ML systems, cybersecurity, and infrastructure automation. When I'm not coding, you'll find me competing in cyber competitions, optimizing my homelab, or experimenting with local AI.

AI Incubator Engineer @ UL Solutions (Starting June 2026)
Joining the AI innovation team to build next-generation solutions

Cybersecurity Engineering Intern @ UL Solutions (Summer 2025)
Gained hands-on experience in security assessment and vulnerability analysis

Student Technology Intern @ Glenview Public School District (Summer 2024)
Worked with IT supervisors at District 34 by aiding and shadowing upper-level network and security managers

Your Internship Journey Begins Here - A comprehensive web platform connecting students with internship opportunities

• Purpose: Streamlines the internship search process by matching students with relevant opportunities based on their skills and interests
• Tech Stack: Flask (Python) Tailwind CSS JavaScript PostgreSQL
• Features:
  • Intelligent job matching algorithm
  • Real-time application tracking
  • Interactive user dashboard
  • Company profile integration
• Etymology: Named from Greek "Arche" (ἀρχή) meaning "beginning" - representing the gateway to career journeys
• Live: archeway.net

Vulnerability Exploitation via Neural Operative Model - AI-powered penetration testing assistant leveraging local LLMs

• Purpose: Automate vulnerability discovery and exploitation guidance while maintaining ethical safeguards and eliminating cloud API dependencies
• Tech Stack: Python Llama 3.3 70B PostgreSQL + pgvector React Docker Metasploit
• Key Features:
  • Local LLM inference (no cloud restrictions or recurring costs)
  • RAG system with 200,000+ CVEs from National Vulnerability Database
  • Tool orchestration (Nmap, Metasploit, SQLMap, Hydra, Searchsploit)
  • Semantic CVE search using vector embeddings
  • Human-in-the-loop approval for all exploit attempts
  • Real-time scan visualization with network topology
• Operational Modes: Stealth, Loud, Reconnaissance-Only, Full Automated
• Safety: Isolated test environment with comprehensive audit logging and mandatory approval workflows
• Research Contribution: Empirical comparison of local vs. cloud LLM performance for security testing tasks
• Model Abliteration (Heretic):
  • Forked Heretic, an open-source LLM abliteration tool using directional ablation with LoRA adapters and Optuna TPE optimization
  • Contributed FP8/NVFP4 quantization support (+59/-21 lines across 5 files), enabling loading of NVIDIA FP8 models that upstream couldn't handle
  • Generated custom datasets for abliteration: 1,000 harmful pentesting prompts and 5,000 enhanced harmless prompts, deployed to Hugging Face (harmful | harmless)
  • Built tooling for automated Parquet conversion, train/test splitting, and Hugging Face repository deployment

Running enterprise-grade infrastructure for learning and experimentation:

• Hardware: Dell PowerEdge R720 server cluster
• Virtualization: Proxmox VE for VM orchestration
• Storage: TrueNAS for centralized data management
• Monitoring: Grafana + Prometheus dashboards
• Networking: VLANs, pfSense firewall, network segmentation
• Services: Self-hosted apps, local AI inference, container orchestration

Active participant in collegiate cybersecurity competitions:

• CDC (Collegiate Defense Competition) - Blue team defense operations (2nd place)
• CTF Challenges - Penetration testing and vulnerability exploitation
• Campus Cyber Club - Regular training and collaborative learning
• CornCon 11 - Networking and knowledge sharing with industry professionals

Languages

AI/ML & Data

DevOps & Infrastructure

Cybersecurity Tools

Databases & Cloud

Monitoring & Observability

• 🎓 Completing dual degrees in CS and AI at Drake University
• 💡 Building AI-powered solutions for real-world problems
• 🔐 Sharpening cybersecurity skills through competitions and CTFs
• 🏗️ Expanding homelab infrastructure with advanced monitoring
• 🤖 Experimenting with local LLM deployments and optimization
• 📚 Exploring RAG systems and vector databases

I'm always interested in collaborating on AI, cybersecurity, or infrastructure projects. Feel free to reach out!

• 💼 LinkedIn: Coleman Pagac
• 📧 Email: [email protected]
• 🌐 Website: colemanpagac.com
• 🤗 Hugging Face: cpagac