This project showcases an AI-driven approach to significantly boost cybersecurity for U.S. critical infrastructure, starting with Water Treatment Systems (WTS).
The core concept is to harness advanced AI ⎯ including LLMs, machine learning algorithms, and Agents employing RAG with human-in-the-loop ⎯ to detect and respond to cyber threats in real-time.
Cyberattacks on U.S. critical infrastructure are a persistent, evolving threat. Malicious actors, including state-sponsored groups from nations like China and Russia, continuously seek to exploit vulnerabilities, jeopardizing essential services and national security. Traditional security often struggles against these sophisticated and rapid attacks.
The US-Infra-Avengers initiative envisions a comprehensive AI-powered safety net for all U.S. critical infrastructure. This system would act as a vigilant guardian, proactively identifying and neutralizing threats before significant harm occurs ⎯ much like a digital "Avengers" for national infrastructure.
The prototype called "Aquaman" initially for WTS, but should be extendable to other sectors, involves:
Note: At the hackathon demo, this project was presented with a slide, the backend (ie., this repository), and an inspirational frontend.
- Real-time Data Analysis: Ingesting and analyzing data from industrial control systems (ICS) and sensors.
- AI-Driven Anomaly Detection: Employing LLMs and ML algorithms to identify unusual patterns and potential threats.
- Intelligent Verification: Utilizing Agents with RAG. These agents access and process info (maintenance logs, operational manuals, threat intelligence, ...) to contextualize anomalies, distinguishing true threats from benign operational deviations.
- Rapid Alerting & Mitigation Support: Providing early, accurate warnings of confirmed attacks for swift human intervention and automated responses.
This "Human-in-the-Loop" approach combines AI's analytical power with human expertise for robust security.
With over a decade of software engineering experience, I am passionate about leveraging technology to solve critical national challenges. This project demonstrates my commitment to applying AI and machine learning concepts to the vital area of national security, and I am eager to contribute my expertise in software architecture and project realization to this field.
I am highly motivated to see the core concepts of this project explored and developed further. I am actively seeking opportunities to collaborate with government agencies, research institutions, or private sector partners on similar challenges. My goal is to dedicate my engineering skills and leadership to a role where I can help build and deploy innovative AI-driven security solutions for our critical infrastructure. With the right team and resources, I believe we can make a significant difference.
Securing 1st Place in the OpenAI Track (National Security) at the SCSP x AGI House Hackathon 2025 was an incredible honor. This achievement was made possible by the vision and support of:
Thank you for fostering innovation against critical national security challenges through technology.
Run each line in your terminal:
conda env create -f environment.yml
conda activate aquaman
conda install -c conda-forge pandas scikit-learn pyod joblibYou should run the notebooks in order, but you may simply take a look at the existing running results from each one.