Muhammad Hamza@Team_HIH

About me

I am an AI student, with a strong focus on building practical AI systems not just experimenting with them. My core interest lies in LLM pipelines, retrieval-augmented generation (RAG), agentic workflows, and deploying optimized ML models that actually work in production. Over the past year, I have built several end-to-end AI projects that solve real problems. One of them is an Agentic RAG Orchestration Engine where I designed a dual-source routing agent using LangGraph that dynamically switches between a vector database and live web search cutting hallucination rates and improving response accuracy by 40%. Another is a Medical Chatbot with a multi-query RAG pipeline that uses the Reciprocal Rank Fusion algorithm across 5 query variations, bringing cold-start latency down from 10 seconds to under 1 second. I am also currently building a voice-enabled AI front desk assistant for a hotel, handling real-time speech recognition, TTS, and multi-turn conversations with context compression. Beyond personal projects, I completed an AI internship at Software Productivity Strategists Inc, where I built a chatbot using IBM Watson Assistant, managed Azure infrastructure, and automated CI/CD pipelines with GitHub Actions and Azure DevOps. On the technical side, I work primarily in Python and am comfortable across the full AI stack - LangChain, LangGraph, TensorFlow, PyTorch, HuggingFace Transformers, FastAPI, Flask, Pinecone, Chroma DB, and Docker. I have worked with both cloud deployments on Azure and local model setups using tools like Ollama and Whisper.

Socials