A conversational AI system for tracking child development (0-3 years), powered by local LLM technology. Combines natural language processing with developmental psychology to provide insights through friendly conversation.
- 💬 Natural Language Interface: Conversational assessment using Mistral 7B LLM
- 📊 Real-time CDDC: Dynamic Child Development Data Chart visualization
- 🎯 Age-Specific Assessment: Tailored developmental milestones tracking
- 🔒 Privacy-First: Local processing with hardware acceleration support
- 🤖 Adaptive Responses: Reinforcement learning for question optimization
- 🌟 Comprehensive Tracking: Seven key developmental domains
Coming Soon! Web interface under development.
# Clone repository
git clone https://github.com/yourusername/child-development-assistant
cd child-development-assistant
# Backend Setup
cd backend
python -m venv venv
source venv/bin/activate # Windows: venv\Scripts\activate
# Install backend dependencies with hardware acceleration
# For macOS (Apple Silicon):
CMAKE_ARGS="-DLLAMA_METAL=on" pip install llama-cpp-python
# For NVIDIA GPU:
CMAKE_ARGS="-DLLAMA_CUBLAS=on" pip install llama-cpp-python
# For CPU only:
pip install llama-cpp-python
pip install -r requirements.txt
# Download LLM model
python scripts/download_model.py
# Frontend Setup
cd ../frontend
npm install
# Run Development Servers
# Terminal 1 (Backend):
cd backend && uvicorn app.main:app --reload --port 8000
# Terminal 2 (Frontend):
cd frontend && npm run dev- Gross Motor: Large movement coordination
- Fine Motor: Hand-eye coordination, manipulation
- Assessment: Movement pattern analysis, milestone tracking
- Receptive: Language comprehension
- Expressive: Speech and communication
- Assessment: Natural language processing, speech pattern analysis
- Problem Solving: Learning and adaptation
- Self-Care: Daily living skills
- Assessment: Task completion analysis, behavioral patterns
- Social Interaction: Peer relationships
- Emotional Expression: Feeling recognition
- Assessment: Sentiment analysis, interaction patterns
- FastAPI for API endpoints
- Mistral 7B (Q3 quantized) for conversation
- Hardware acceleration support (Metal/CUDA/CPU)
- SQLAlchemy for data persistence
- Spacy for NLP processing
- React 18 with TypeScript
- Material-UI for components
- Chart.js for visualization
- Vite for development
- Local LLM inference
- Reinforcement learning for question selection
- Developmental milestone mapping
- Sentiment analysis pipeline
# Hardware-specific optimizations
LLM_CONFIG = {
"model_path": "models/mistral-7b-instruct-v0.2.Q3_K_S.gguf",
"n_ctx": 2048, # Adjust based on available memory
"n_threads": 8,
# Hardware acceleration settings:
# Apple Silicon: n_gpu_layers=-1
# NVIDIA GPU: n_gpu_layers=35
# CPU only: n_gpu_layers=0
"n_gpu_layers": -1,
"n_batch": 512, # Adjust based on GPU memory
# Optional hardware-specific settings
"use_mmap": True,
"use_mlock": True
}/api/assessment/start: Initialize assessment session/api/assessment/process: Process conversation/api/speech/*: Speech-to-text and text-to-speech/api/auth/*: Authentication endpoints
Q: System Requirements? A: Minimum Requirements:
- Any modern CPU (Intel/AMD/ARM)
- 8GB RAM
- 5GB storage
- Python 3.11+
- Node.js 18+
Recommended Hardware:
- Apple Silicon Mac (M1/M2/M3) OR
- NVIDIA GPU (8GB VRAM) OR
- Modern CPU (8+ cores)
- 16GB RAM
Performance Tiers:
- Best: Apple Silicon Mac or NVIDIA RTX 3060+ GPU
- Good: NVIDIA GTX 1660+ GPU
- Basic: Modern multi-core CPU
Q: Is this a diagnostic tool? A: No, this is a development tracking assistant. Consult healthcare professionals for diagnosis.
Q: Data Privacy? A: All processing is local. No data leaves your device.
Q: Performance on different hardware? A: The system adapts to available hardware:
- Apple Silicon: Fastest, most energy efficient
- NVIDIA GPU: Similar performance to Apple Silicon
- CPU only: Slower but fully functional
See CONTRIBUTING.md for guidelines.
MIT License - See LICENSE for details.
Made with ❤️ for families and developers alike