Built on Next.js 14 with Groq's Llama-3.3-70b for ultra-fast inference, it streams AI responses in real time while dynamically injecting interactive React components (ProjectCards, SkillPanels, ContactForms) directly inside the chat. Key engineering: SSE-based streaming from the edge, hidden context routing per navigation tab, component parsing mid-stream, and immutable state management for a race-condition-free UI. A portfolio that acts as a live AI demo, visitors don't read about what I build, they experience it.