AI & Security Engineer | System Architect | Tool Maker

Welcome to my digital workspace! I specialize in building Autonomous AI Systems, advanced RAG architectures, and Cybersecurity solutions. I am passionate about bridging the gap between cutting-edge AI models (LLMs) and real-world enterprise infrastructure, as well as empowering local developer communities.

• Agentic AI & LLM Orchestration: Designing fallback routing systems, dynamic context windows, and multi-agent workflows.
• Advanced RAG Systems: Leveraging Vector Databases and Knowledge Graphs for hallucination-free AI memories.
• Automated Security: Building autonomous agents for CVE analysis, threat intelligence, and auto-patching pipelines.
• AI Framework Development: Architecting accessible, native-language machine learning tools for developers.

• AI/ML: Qwen, Llama, HuggingFace, Custom ML Pipelines, Prompt Engineering, Agentic Frameworks.
• Backend: Python (FastAPI), Node.js, C++.
• Databases: Pinecone (Vector DB), Neo4j (Graph DB), PostgreSQL, Firebase.
• Infrastructure: Docker, Linux (NUC Baremetal), Cloudflare AI, Groq, NVIDIA NIM.

An Enterprise-grade No-Code/Low-Code LLM Fine-tuning UI.

• Publication:
• Capabilities: 1-click LLM fine-tuning via an intuitive Gradio interface, empowering developers to align models without complex environment setups.
• Architecture: Smart Hybrid CPU/GPU detection. Auto-loads Unsloth kernels for 2x faster 4-bit training on GPUs, with graceful fallbacks to native HuggingFace on CPUs.
• Deployment: Fully Dockerized (docker-compose) for instant cloud or on-premise execution.

The first production-ready Machine Learning framework with a 100% native Vietnamese API.

• Publication:
• Capabilities: Comprehensive ML toolkit featuring native Vietnamese NLP (Underthesea integration), AutoML, Cross-validation, and Neural Networks.
• Architecture: Modular design with built-in Pipeline support, custom CLI (vai), and Docker-ready deployment.
• Mission: Lowering the barrier to entry for Vietnamese developers by providing highly accessible, native-language AI tools and documentation.

An enterprise-grade Autonomous AI Security Agent.

• Capabilities: Real-time CVE harvesting, automated vulnerability scanning, and autonomous code patching.
• Architecture: 4-Layer LLM Fallback Routing (NVIDIA -> Groq -> Cloudflare -> Local).
• Memory: Powered by Pinecone and Neo4j for deep contextual threat intelligence.

📫 How to reach me: Connect with me on LinkedIn