I built a "Sentient" Portfolio using Llama-3 and React (Generative UI)

I got bored of static portfolios. So I built one that thinks.

Instead of hardcoding a layout, I hooked up my frontend to Llama-3-70b (via Groq). It analyzes the visitor's intent and physically re-architects the grid using Framer Motion.

🧠 How it works

1. User Prompt: "I want to hire you."
2. LLM Inference: Logic runs on Netlify Functions (<200ms latency via Groq LPU).
3. State Mutation: The AI returns a JSON config that changes the React components from "Explorer Mode" to "Resume Mode."

🛡️ The "Janitor" Guardrail

I added a safety layer. If you ask: "I want to hire you as a Janitor," the AI detects the context, refuses the prompt with a witty retort, and resets the UI.

🛠️ The Stack

• Brain: Llama-3-70b (Groq)
• Frontend: React + Framer Motion + Bento Grids
• Backend: Netlify Functions

🔗 Try it here

👉 Click here to launch the AI Portfolio

🐦 Follow the Journey

I'm documenting the build on Twitter:

Let me know if you can break the prompt injection! 👇