Jiabao Chau | AI & Cybersecurity

THE
ADVERSARIAL
MINDSET.

Security & Vulnerability Focus

Successfully bypassed Level 7 of Lakera Gandalf’s security challenges, demonstrating a critical ability to identify prompt injection vulnerabilities and LLM exploits.

Technology isn't just a tool; it's a boundary to be pushed. Since the 2020 transition to online learning, I have approached AI as a puzzle to be solved and a system to be rigorously tested.

While many focus on building for convenience, I focus on building for resilience. My Philosophy is rooted in "learning by doing", from engineering AI-powered moderation tools to probing complex server structures. The best way to secure a system is to understand exactly how it breaks.

ENGLISH / Fluent MANDARIN / Fluent CANTONESE / Fluent

SYSTEM ARCHITECTURE

Technical Arsenal

Development & Logic

01

■ Advanced Python development for AI-driven Discord architecture and security tooling.
■ Creation of AI-powered QoL programs including automated screenshot redaction and media optimization pipelines.
■ Proficient across diverse IDEs/environments: VSCode, Pycharm, Kiro, Antigravity, Replit, Windows scripting.

AI Research & Inference

02

■ Deployment of local LLM inference providers (Ollama, llama.cpp) for sovereign intelligence applications.
■ Hands-on experience with LLM jailbreaking, prompt injection diagnostics, and AI security testing.
■ Trained and deployed custom computer vision models for automated community moderation.

SELECT INTELLIGENCE

INFRASTRUCTURE SEC 01

AEGIS

A hardened Discord staff management suite designed for high-stakes community administration. Featuring Argon2id password hashing, Just-In-Time (JIT) privilege elevation, and periodic scans to automatically detect and neutralize rogue administrative access.

Zero-Trust Architecture Audit Logging

RECURSIVE AGENT 02

A.U.R.A. MK-V

A fully agentic AI web search system deployed within Discord. Utilizes a CLI-based automated browser to recursively gather, research, verify, and compile multi-source deep-web data into comprehensive intelligence reports to answer complex user queries.

Autonomous Execution Data Compilation

DATA SYNTHESIS 03

A.U.R.A. OSINT-2.1

An autonomous system built for OSINT (Open Source Intelligence) gathering. Leverages a multi-pipeline RAG architecture and local LLMs (Ollama) to scrape, cache, and apply agentic reasoning to extract actionable intelligence from massive Discord channel datasets.

Local LLM Engine RAG Pipeline

SECURITY AUDIT 04

VANTAGE Probe

A specialized security loopholes auditor for complex Discord role hierarchies. Employs multi-agent collaboration powered by local LLMs to dynamically identify and report privilege escalation loopholes and RBAC vulnerabilities.

Multi-Agent Workflow RBAC Security

INITIATE
COMMUNICATION.

Open to exploring opportunities in AI safety, red-teaming, and agentic system architecture. High-fidelity reporting and documentation guaranteed.

Direct Email

superprofessionalemail@gmail.com

Direct Line

(510) 656-5711

Location Data

41800 Blacow Rd

References

Michael J Schmidt

Close Friend

frankielow9@gmail.com | 801-942-8185

Ronald R Stumpf

Family Friend

zoey_lind1983@hotmail.com | 734-432-3518

David E Kramer