Abdullah Naeem

I’m an Artificial Intelligence undergraduate focused on machine learning, computer vision, and agentic AI systems. I build practical, demo-to-deployment style solutions—especially where AI research translates into measurable real-world impact.

I’ve developed an AI-powered surveillance anomaly detection system (SurveilX) using CNN and CLIP, including visual prompting for natural-language scenario search and a real-time backend with Uvicorn plus an interactive React dashboard. I also built a multi-agent disaster response “Crisis Simulator” coordinating drones and emergency vehicles, using ReAct and Reflexion frameworks for planning and decision-making.

Beyond anomaly detection, I design agent workflows and automation: Voyant uses an agentic LangChain workflow with real-time retrieval and reasoning to generate personalized travel itineraries. ConstructGuard applies CNN models and PyTorch to enable real-time safety compliance detection for construction sites, training detectors for helmets and protective equipment.

I’m equally interested in learning systems and verification pipelines. I applied reinforcement learning to optimize crop yield in a Maize Production Optimizer project, and I built a media provenance app (VeriVision) using perceptual hashing plus deep feature similarity with ResNet50 embeddings and FAISS indexing. I’ve also created an LLM-powered social media automation workflow using n8n, Gemini API, and X API.