# FAT&GCR

- **Event:** [Meta Global AI Developer Hackathon](https://cerebralvalley.ai/e/meta-ai-developer-hackathon)
- **When:** Oct 26, 12:00 PM – Nov 4, 9:00 PM (EST)
- **Where:** Online
- **Team:** [Felix Ahmet Tapia-Valentin](https://cerebralvalley.ai/u/WatusPulpo)
- **GitHub:** https://github.com/ftapi/Meta_llama-Powered-AI-Cybersecurity-Intelligence-and-RAG-Platform
- **Demo video:** https://youtu.be/n3X2qJrCw5I
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/5

The solution will integrate Meta Llama, Ollama, Open WebUI, Qdrant Vector Database, Python, Docker, and NVIDIA GPUs** to create an intelligent cybersecurity assistant capable of analyzing security documentation, processing threat intelligence, and supporting Security Operations Center (SOC) activities.

The platform will operate on a self-hosted Ubuntu Linux infrastructure, providing greater control over sensitive enterprise security information without requiring external cloud-based AI inference.

What the Project Will Do

- AI-Powered Cybersecurity Analysis: Using Meta Llama models to analyze vulnerabilities, security incidents, attack patterns, and potential threats.
- Intelligent Knowledge Retrieval: Implement RAG using Qdrant to retrieve information from cybersecurity documents, technical manuals, security policies, and threat intelligence repositories.
- Security Telemetry Integration: Ingest and process security events from firewalls, SIEM platforms, endpoint security tools, intrusion detection systems, and Windows/Linux system logs.
- Automated Incident Investigation: Correlate relevant security information, summarize suspicious activities, and assist analysts with incident triage and response recommendations.
- Natural Language Security Assistant: Enable security professionals to ask questions about vulnerabilities, security events, compliance requirements, and infrastructure configurations through Open WebUI.
- Automated Data Processing: Develop Python-based ingestion pipelines for document extraction, text chunking, embedding generation, vector indexing, and scheduled knowledge-base updates.
- GPU-Accelerated AI Inference: Optimize local Meta Llama execution using NVIDIA RTX 3090 and Tesla V100 GPUs, subject to model compatibility and available GPU memory.
- Private AI Infrastructure: Maintain control of confidential data, AI processing, knowledge repositories, and security telemetry within the organization's infrastructure.

Technologies and Techniques

Technologies: Meta Llama, Ollama, Open WebUI, Qdrant, Python, Bash, Docker, Ubuntu Linux, NVIDIA CUDA, and REST APs.

AI Techniques: Retrieval-Augmented Generation (RAG), semantic search, vector embeddings, prompt engineering, document chunking, LLM inference, security event correlation, and workflow automation.

Expected Outcome

Deliver a scalable, self-hosted AI cybersecurity platform that transforms enterprise security documentation and operational telemetry into actionable intelligence.

The solution will help **accelerate threat investigations, reduce manual analysis, improve access to institutional knowledge, and enhance SOC analyst productivity**, establishing a foundation for more advanced AI-assisted security operations .

## More from Meta Global AI Developer Hackathon

- [Halimi](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/2)
- [md iqbal hossain](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3)
- [Jass's team](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4)
- [OPERACIJA DVINA](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/6)
- [_123olalekan](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/7)
- [solopap](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/8)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/5. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt