These projects represent the cutting edge of my recent work in AI, automation, and intelligent interfaces.
Avatar Hub
A modular, self-hosted avatar framework that wires together automatic speech recognition, LLMs, speech-to-text, and video rendering. A complete end-to-end pipeline for conversational AI. Optimized with CUDA and ONNX.
Caricature Maker Kiosk
An interactive, AI-powered kiosk using Stable Diffusion and MediaPipe for real-time, automated portrait generation.
AI Chatbot Agents
A multi-agent AI architecture platform in which specialized agents (Diagnostics, Billing, RAG) can work together within a customer service interface.
Hard Hat Detector Gate
A real-time PPE compliance monitoring system using YOLOv8 for industrial safety automation.
In-Browser LLM Assistant
A privacy-first, zero-server personal finance assistant running entirely in the browser via WebAssembly (Wllama).
Vision Attribute Pipeline
A real-time computer vision pipeline for clothing attribute extraction to attract people towards a conference booth (“hey, you in the red shirt!”), using YOLOv8 and OpenVINO.
Campaign QA Suite
An automated testing tool using Playwright and an LLM to validate live marketing campaigns against visual and functional requirements.
Product Classification
A fine-grained visual classification API for automated hardware identification via smartphone photography.