At Nexonic, I work on AI modules in a production development environment, including a RAG-based sales assistant, a computer vision pipeline, and autonomous agent systems. I developed the sales assistant with hybrid retrieval and re-ranking, and engineered a supplier negotiation agent integrated with WeCom.
I also built a real-time vehicle detection pipeline for surveillance-style object detection tasks. My work at Nexonic combines retrieval, computer vision, and automation.
As a Research Assistant at MNS University of Agriculture, Multan, I prepared agricultural image datasets and trained segmentation and classification models for plant disease detection. The models achieved 95%+ accuracy.
I built Zarai Radar, an AI crop-risk intelligence platform that combines agronomic modeling with RAG to deliver explainable, location- and growth-stage-aware advisories. I’ve also developed RAG and computer vision projects, including a biomedical literature system and a lung CT scan classifier.
