At Systems Limited, I contributed to AI-powered applications using LangChain, Retrieval-Augmented Generation (RAG), and Model Context Protocol (MCP). I also helped migrate AI workflows into agentic systems and implemented backend services with FastAPI.
For my automatic bill generation project, I fine-tuned YOLOv8 to identify food items in tray photographs and generate itemized bills, with results stored in Supabase. I also engineered a CI/CD pipeline using GitHub Actions, Docker, and AWS.
On a project for real-time Pakistani vehicle counting, I curated and labeled a custom dataset of local vehicle classes, then fine-tuned YOLOv8 for detection and counting. The work addressed vehicle types underrepresented in existing training data.
I'm a final-year BS Artificial Intelligence student at FAST NUCES, Karachi. My projects also include reproducing a research paper's image-based shape inference system with PyTorch, CNNs, Spatial Transformer Networks, and MLPs.
