$whoami
I'm a cybersecurity researcher transitioning into AI security research, with professional experience in cloud security, offensive security, and application security. Currently studying at the Judy Genshaft Honors College at the University of South Florida in the Bellini College of Cybersecurity and AI.
- Wells Fargo (Incoming Full-Time): Built the company's first Terraform IAM testing modules for GCP & Azure, developed AI-assisted security automation POC.
- Honeywell Aerospace: Red-teamed flight-critical systems, found RCE vulnerabilities, presented findings to CISO.
- JPMorgan Chase: Application security analysis, ML-based fraud detection, SSDLC hardening.
I actively compete in national CTFs with USF CyberHerd (Top 5 nationally) and founded the USF Boxing Club, which scaled to over 1,300 members-the largest combat-sports organization on campus.
AI Security Research
My transition into AI security is motivated by bringing traditional penetration testing methodology to AI systems-systematically probing where vulnerabilities hide in interactions between ML components and infrastructure.
Current Research Areas:
- Offensive AI System Evaluation: Applying pentesting techniques (session manipulation, API fuzzing, privilege escalation) to discover vulnerabilities in model APIs, agent frameworks, and ML infrastructure
- Composite Attack Chains: Researching how traditional security vulnerabilities chain with AI-specific weaknesses (e.g., rate limiting + prompt injection to infer model architecture)
- LLM Jailbreak Analysis: Systematic testing of open-source models (Llama 3, Mistral) with telemetry capture to identify predictable failure patterns
- Adversarial ML Security: Exploring robustness of computer vision models against adversarial attacks (FGSM) in malware detection contexts
Experience
Notable CTF Wins
- 1st – EC-Council CTF (Captain)
- 1st – MetaCTF 2025
- 2nd – NASA × HackSpaceCon RFCTF
- 2nd – CPTC Southeast Regionals 2025
- 3rd – SECDCC 2025
- Top 5 – Hivestorm 2024
- Finalist – Splunk Boss of the SOC
- Finalist – Palo Alto Secure the Future
Key Achievements
- Led red-team engagement on 10,000+ internal IPs at Honeywell Aerospace
- Presented aerospace-hacking findings directly to Honeywell's CISO
- Built Wells Fargo's first GCP IAM Terraform testing modules to detect misconfigurations
- Led USF CyberHerd-a Top 5 national team-to multiple CTF wins
- Founded & scaled USF Boxing Club to 1,300+ members (largest in campus history)
- Named 2024-25 USF Sport Club Leader of the Year
Projects
Jailbreak Telemetry & Attack Surface Mapping
Python framework automating jailbreak testing on open-source LLMs (Llama 3, Mistral) with granular telemetry capture-token distributions, latency patterns, resource metrics. Developing interactive visualizations of attack surfaces to identify predictable failure modes in smaller fine-tuned models vs. large base models.
Composite Attack Chain Research (Ongoing)
Investigating how traditional security vulnerabilities chain with AI-specific weaknesses to create novel attack vectors. Examples: exploiting rate limiting + prompt injection to infer model architecture; chaining authentication bypass + agent tool poisoning to compromise multi-agent systems. Researching whether composite attacks follow predictable patterns that can be systematically enumerated and defended against.
AI Malware Vision Classifier
End-to-end deep learning pipeline converting malware binaries into grayscale images for classification. Lightweight CNN achieving 98.90% test accuracy on Malimg dataset. Integrated GradCAM explainability for model interpretation and FGSM adversarial robustness analysis to study failure modes under attack.
Anbu Full-Stack CSPM
Cloud Security Posture Manager detecting and remediating misconfigurations across AWS/GCP with focus on identity misconfiguration detection and visualization. Built for at-scale security posture management in multi-cloud environments.
Secure Cloud CI/CD Pipeline
Production-grade DevSecOps reference implementation with automated build, scan, deploy, and self-healing for containerized workloads on AWS (cloud-agnostic configurable). Integrates Trivy scanning, policy-as-code enforcement, and automated rollback.
Azure SIEM for RDP Attack Detection
Dynamic detection system using Azure Sentinel and KQL to monitor global RDP brute-force attempts with automated remediation workflows. Real-time threat intelligence integration and geolocation-based alerting.
Explore more projects on GitHub →
Certifications
In Progress: OSCP, AWS Security Specialty, Terraform Associate