Available for AI security roles
Chima Ukachukwu. AI security · LLM red teaming · SOC operations.
Replies within 24 hours, weekdays.
Three things I built. Each one names what it demonstrates.
Interactive · client-side
My jailbreak taxonomy as a working classifier. Ten categories, each defined by mechanism rather than phrasing, with the matched pattern shown for every hit, because a detection you can't explain isn't a detection.
Demonstrates taxonomy design, detection logic, explainable output
Case study
Three chained agents that extract contract clauses, red-team each one for hidden risk, then draft the counter-proposal. Built in 48 hours.
Demonstrates multi-agent architecture, adversarial scoring
Writing
Why traditional SIEM primitives partially fail against prompt injection, three places rule-writing can actually start, and what stays hard.
Demonstrates detection engineering, a point of view
Most AI security people came from machine learning. Most SOC analysts have never tested a model. I work from both sides, and the transfer runs in both directions.
Threat modeling, structured methodology and a security mindset move cleanly between traditional infrastructure and frontier AI systems. Knowing how detection fails in a SIEM tells you what an AI security control will miss. Knowing how a model fails tells you what is worth logging in the first place.
I started in IT support and systems administration in 2014, running helpdesk, LAN/WAN and endpoint security before moving into security operations and then AI security. The infrastructure years are not background detail. Knowing how systems fail in production is what makes model failures legible.
On the offensive side I run adversarial testing against frontier large language models: jailbreak analysis, prompt injection assessment and misuse evaluation. I publish the sanitized methodology on GitHub rather than the working exploits. On the defensive side I worked a live enterprise SOC at Hobby Lobby Corporate IS, deploying MISP threat intelligence automation in Python and monitoring events with Splunk.
I hold an M.S. in Cybersecurity (GPA 3.7 / 4.0) from Oklahoma City University.
Plus Azure AI Fundamentals, TryHackMe paths, TCM Security, and Forage simulations (Mastercard, PwC, AIG).
A shell over the same classifier that runs in the Lab. Type scan <prompt> to have it dissect a prompt injection, or help for the rest.
██████╗██╗ ██╗██╗███╗ ███╗ █████╗
██╔════╝██║ ██║██║████╗ ████║██╔══██╗
██║ ███████║██║██╔████╔██║███████║
██║ ██╔══██║██║██║╚██╔╝██║██╔══██║
╚██████╗██║ ██║██║██║ ╚═╝ ██║██║ ██║
╚═════╝╚═╝ ╚═╝╚═╝╚═╝ ╚═╝╚═╝ ╚═╝
🔐 AI Security Analyst & Red Teamer
M.S. Cybersecurity (GPA 3.7) | Industry Certs + Ongoing Training | A Decade in IT
Type help to see available commands, lab for the interactive exhibits, or whoami to start.
Open to AI security, AI red team, SOC analyst and GRC roles, entry to mid level. U.S.-remote or Oklahoma City.
Useful things to send me: an AI system you want tested, a SOC role that touches model security, or a question about anything in the Lab. I reply to every message that isn't automated.