Neural Shadow: AI-Driven Cloud Security Orchestrator
💡 InspirationMy: journey to creating Neural Shadow didn't start in a computer lab—it started in the patrol car and the machine shop. Before my transition into IT, I spent five years as a Texas Law Enforcement Officer and served as a Lead Mechanical Diesel Technician.I realized that cybersecurity is simply the digital evolution of these roles: Investigation, Diagnostics, and Protection. I built Neural Shadow to automate the "patrol" of cloud environments, using the same root-cause diagnostic mindset I used for diesel engines to identify complex cloud misconfigurations.
⚙️ What it does Neural Shadow is a "Security Orchestrator" that bridges the gap between finding a vulnerability and understanding its risk.Automated Scanning: It identifies high-risk exposures like TOKEN_LEAKAGE and PUBLIC_BUCKETS.AI Reasoning: Using GPT-4o-mini, it performs a "Kill Chain Analysis" on every finding.Risk Quantification: We calculate threat priority using a weighted logic model:$$Risk = (Threat \times Vulnerability) \times Impact$$
🛠️ How we built it The project is built on a modular, secure stack:Frontend: Streamlit for a real-time, responsive mission control.Backend: Custom Python logic (the "OWEN" engine) for GCP credential validation.Intelligence: OpenAI API for natural language threat analysis.Security: Implemented a Zero-Trust posture using TOML-based secrets management to protect GCP and OpenAI keys.
🚧 Challenges we overcameAs a student at South University, the primary challenge was balancing a rigorous Bachelor of Applied Science curriculum with the rapid development cycle of a hackathon. I had to ensure my code didn't just "work," but followed the strict software engineering principles I'm learning in my classes—specifically regarding secure CI/CD pipelines and the Principle of Least Privilege for service accounts.
🏆 Accomplishments that we're proud of I successfully deployed a live, functional security agent that can communicate across two major cloud ecosystems (OpenAI and GCP) while maintaining a perfect security posture. Seeing the agent correctly identify an information disclosure similar to reports I've filed with the U.S. Department of Defense was a major milestone
đź“š What we learned I deepened my understanding of Identity and Access Management (IAM) and the critical importance of "Identity as the New Perimeter." I also mastered the process of automated cloud deployment via GitHub, learning how to manage complex dependencies in a requirements.txt environment.
Log in or sign up for Devpost to join the conversation.