╔╦╗╦ ╦╦ ╔╦╗╦╔╦╗╔═╗╔╦╗╔═╗╦  
║║║║ ║║  ║ ║║║║║ ║  ║║╠═╣║  
╩ ╩╚═╝╩═╝╩ ╩╩ ╩ ╚═╝═╩╝╩ ╩╩═╝
        AI  AGENT             

"What if AI could look at an image and talk about it like a human?" That's exactly what this does.

---

🧠 The Idea

Most AI is either good at text or good at images — rarely both.

This agent bridges that gap. Upload any image, ask any question, and watch it think through vision + language together — like a human would.

      You upload an image  +  You ask a question
               │                      │
               ▼                      ▼
         ┌──────────┐          ┌────────────┐
         │   BLIP   │          │  Your Text │
         │ (Vision) │          │   Query    │
         └────┬─────┘          └─────┬──────┘
              │                      │
              ▼                      ▼
         📝 Caption ────────▶ 🧩 Combined Prompt
                                      │
                                      ▼
                              ┌───────────────┐
                              │  LLaMA 3.3    │
                              │  via Groq ⚡  │
                              └───────┬───────┘
                                      │
                                      ▼
                              💬 Intelligent Answer

---

✨ What Makes It Special

🖼️ Vision Understanding Upload any image — photos, diagrams, screenshots, charts — and the agent understands what's in it before you even ask.	💬 Natural Language Q&A Ask follow-up questions in plain English. No special syntax. No prompting tricks. Just talk.
⚡ Groq-Powered Speed LLaMA 3.3 70B running on Groq's LPU chips means answers arrive in under 2 seconds — not 20.	☁️ Live & Deployed Not just a local demo. Actually deployed on Streamlit Cloud and accessible from anywhere in the world.

---

🏗️ Tech Stack

┌─────────────────────────────────────────────────────────────┐
│                                                             │
│   Frontend  ──────────────────────────  Streamlit           │
│   Vision    ──────────────────────────  BLIP (Salesforce)   │
│   Language  ──────────────────────────  LLaMA 3.3 70B       │
│   Inference ──────────────────────────  Groq API ⚡         │
│   Language  ──────────────────────────  Python 3.10+        │
│   Hosting   ──────────────────────────  Streamlit Cloud     │
│                                                             │
└─────────────────────────────────────────────────────────────┘

Layer	Technology	Why
🖼️ Image Captioning	BLIP (Salesforce)	State-of-the-art vision-language model
🤖 LLM	LLaMA 3.3 70B	Open-source, powerful, free
⚡ Inference	Groq API	10x faster than standard GPU inference
🎨 UI	Streamlit	Fast to build, beautiful to use
☁️ Deploy	Streamlit Cloud	Free, instant, shareable

---

📂 Project Structure

multimodal-agent/
│
├── 📄 app.py                 ← main streamlit app + UI
│
├── 📂 model/
│   ├── 🖼️  blip.py           ← image captioning pipeline
│   └── 🤖  llm.py            ← groq + llama inference
│
├── 📂 utils/
│   └── 🧩  prompt.py         ← prompt builder (caption + query)
│
├── 📄 requirements.txt       ← all dependencies
├── 📄 .env                   ← GROQ_API_KEY (don't commit!)
└── 📄 README.md

---

🚀 Run Locally

# 1. Clone the repo
git clone https://github.com/SHALINISAURAV/Multimodal_Agent.git
cd Multimodal_Agent

# 2. Create virtual environment
python3 -m venv venv
source venv/bin/activate        # Mac/Linux
# venv\Scripts\activate         # Windows

# 3. Install dependencies
pip install -r requirements.txt

# 4. Set your API key
echo "GROQ_API_KEY=your_api_key_here" > .env

# 5. Launch 🚀
streamlit run app.py

🔑 Get your free Groq API key → console.groq.com

---

🔐 Environment Variables

GROQ_API_KEY=your_groq_api_key_here

For Streamlit Cloud deployment — add this in Settings → Secrets (not in .env file).

---

🌟 Roadmap

• 🖼️ Image upload + captioning
• 💬 Text Q&A on images
• ⚡ Groq-powered fast inference
• ☁️ Streamlit Cloud deployment
• 🧠 Chat history & memory
• 🎤 Voice input support
• 👁️ Upgrade to LLaVA / Gemini Vision
• 📊 Multi-image comparison
• 🌍 Multilingual support

---

👩‍💻 Author

Shalini Saurav

AI Developer · Builder

Building intelligent systems that understand the world — one model at a time.

---

⭐ If this helped you, star the repo — it keeps the motivation alive!

"Building intelligent systems that understand the world like humans."