Abdul Mueed Khan

I’m a Software and AI/ML Engineer with a strong Computer Science foundation, focused on turning LLM and ML ideas into working, maintainable systems. I’ve built end-to-end pipelines that span data ingestion, modeling, and production deployment.

In my internships, I engineered a financial data parsing system with intelligent column-type detection and optimized indexing, then created a transaction reconciliation engine for the Subset Sum Problem using progressive approaches (exact matching, ML-based reconciliation, genetic algorithms, and fuzzy matching). I also designed a multi-agent reconciliation system that replaces static rules with ML-driven threshold selection and includes explainable AI audit trails.

I’ve also architected a serverless ML pipeline with full CI/CD integration using GitHub Actions and Hopsworks to manage feature storage and automate hourly ETL ingestion. For forecasting, I developed a dual-band Multi-Horizon LSTM model for PM2.5/PM10 with log-scaling and outlier removal, achieving strong predictive performance, and I containerized inference with Docker and deployed via FastAPI for low-latency results.

My final-year work and projects reflect my emphasis on reliable AI: an LLM-driven recruitment platform with multi-stage interview agents, resume parsing, and evaluation; a WHO clinical guidelines RAG system using FAISS with intent-aware hybrid re-ranking and a medical safety layer with citation tracking; and an end-to-end text classification pipeline deployed through FastAPI.