Team_OG

🚀 Team_OG Project Description 💡 Inspiration

Modern healthcare solutions often overlook traditional knowledge systems like Ayurveda, which are deeply rooted in preventive and holistic care. At the same time, accessing reliable health information—especially in regional languages—remains difficult.

We were inspired to build an AI-powered Ayurvedic health assistant that:

bridges modern AI with traditional Ayurvedic wisdom supports multilingual users across India delivers personalized, holistic health insights 🧠 What it does

Our system is a multilingual AI-powered Ayurvedic health assistant using RAG (Retrieval-Augmented Generation).

It enables users to:

Ask health-related questions in English, Hindi, Tamil, Telugu Get context-aware answers grounded in Ayurvedic and medical knowledge Receive personalized lifestyle, diet, and herbal recommendations

Key features:

🌿 Integration of Ayurvedic concepts (Doshas, Prakriti, Herbs, Remedies) 📚 Knowledge base combining structured medical + Ayurvedic datasets and PDFs 🌐 Multilingual interaction 🧠 Semantic search using vector embeddings 💬 Conversational AI interface 📊 Personalized health insights dashboard ⚙️ How we built it 🔹 Data Pipeline Ingested structured CSV (including Ayurvedic attributes like Doshas, Herbs, Diet) Processed PDF sources containing Ayurvedic formulations and remedies Cleaned multilingual noisy data and structured it for AI use 🔹 Embeddings & Search Generated embeddings using Databricks models Built Vector Search index for semantic retrieval of Ayurvedic + medical knowledge 🔹 RAG Pipeline User query → embedding → top-k retrieval Context passed to LLM for accurate, grounded responses 🔹 Multilingual Layer Language detection Translation pipeline for Indian languages Response returned in user’s native language 🔹 Frontend Interactive UI using Streamlit User profile inputs (diet, stress, lifestyle) Displays Ayurvedic recommendations + health insights ⚠️ Challenges we ran into Cleaning and structuring multilingual Ayurvedic texts Preserving context across complex medical + traditional knowledge Handling mixed-language inputs (regional + English) Ensuring accurate retrieval for domain-specific Ayurvedic concepts Balancing latency vs response quality 🏆 Accomplishments that we're proud of Built a complete end-to-end multilingual RAG system Successfully integrated Ayurvedic knowledge into AI responses Designed a system that combines: modern ML semantic search traditional health systems Created a personalized health assistant with lifestyle + herbal insights Achieved fast, relevant responses (<2–3 sec) 📚 What we learned Data quality is more important than model complexity Structuring domain-specific data (like Ayurveda) is critical Multilingual AI systems require careful translation + normalization RAG pipelines depend heavily on: chunking context quality Learned how to build real-world AI systems, not just models 🔮 What's next for Team_OG 🔊 Add voice-based interaction (vernacular speech) 🌿 Expand deeper into Ayurvedic diagnosis & dosha analysis 🧠 Improve personalization using user history and health patterns 🌍 Support more regional languages 📱 Deploy as a mobile-first application 📊 Build advanced Ayurvedic health analytics dashboard

Built With

• datalake
• particle