Abdullah Usman

I'm building an automated time-series forecasting assistant that turns natural-language requests into structured forecasting specifications, validates inputs, engineers features, and benchmarks models with rolling-origin evaluation.

Through internships at Punjab Information Technology Board and Blunder Bot Technologies, I support AI/ML solutions for public-sector transformation and build LLM-powered chatbot and messaging automation workflows from design through deployment. At Swift Solutions, I trained and fine-tuned LLMs, evaluated output quality, and tested LLM workflows for reliability.

I've also built RAHATAI, an Urdu-English RAG disaster-response system using Llama-2, LangChain, and FAISS; an AR indoor navigation system; and a React Native streaming platform with Node.js and PostgreSQL.