← All projects
Personal project2025 - Present

AI-Powered Assistant System

Smart Automation & Communication with AI

AI-based assistant system built on local large language models (LLMs) and vision-language models (VLMs), designed to enhance automation, communication, and intelligent interaction across platforms and real-time image processing systems.

Local LLMOpen APIVision-Language

The work

Key features

  • Local deployment of LLMs (Ollama, Llama.cpp) for fast and private AI processing
  • Open API for connecting AI capabilities to external applications
  • Real-time object identification using camera and VLMs
  • Contextual AI replies via WhatsApp for assistance and inquiries

The impact

Results

  • 100% data privacy with fully local processing
  • Rapid response times for user queries
  • Automated visual inspection capabilities
Technical approach
  • Deployed using Ollama and Llama.cpp for efficient local inference
  • Integration with vision-language models for image understanding
  • RESTful API design for easy integration with existing systems
  • WhatsApp integration for conversational AI
  • Real-time camera feed processing with object detection
  • Resource optimization for deployment on standard hardware
Challenges
  • Optimizing model performance on limited hardware resources
  • Integrating multiple AI models in a cohesive system
  • Ensuring data privacy and security with local processing