shamiimuhammad/NexGen-Support-Hero

🤖 NexGen Support Hero

Industrial-Grade RAG with Identity Anchoring & Persistent Memory

An agentic customer support system built with LangChain, Groq (LPU), and Streamlit. Designed to handle technical inquiries while maintaining a strict professional persona.

🚀 Key Features

• —Identity Anchoring: Hard-coded system prompts prevent "persona injection" (e.g., refusing to act like a pirate or surfer).
• —Stateful Persistence: Uses SQLite to ensure chat history survives browser refreshes.
• —Industrial RAG: Optimized retrieval using FAISS and Sentence-Transformers.
• —Observability: Integrated with LangSmith for full trace auditing.
• —Dockerized: Fully containerized for consistent deployment across environments.

🛠️ Tech Stack

• —Brain: Groq Llama-3 (via LangChain)
• —Memory: SQLite & Streamlit Session State
• —Vector DB: FAISS (CPU)
• —Package Manager: uv
• —DevOps: Docker (Slim-Debian base)

📦 Local Setup (Docker)

1. 1.Clone the repo.
2. 2.Add your .env file with GROQ_API_KEY.
3. 3.Build the image:

   docker build -t support-hero .