Muhammad Murtaza

Senior Python AI Engineer with 4+ years of production experience building backend systems and APIs for AI-powered applications. I specialize in LLM orchestration (LangChain, LangGraph, LlamaIndex), RAG pipeline design, and multi-agent architectures that automate real business workflows at scale.

I've shipped systems serving 50K+ monthly users, processing 2M+ document pipelines, and handling 100K+ daily API requests at sub-200ms latency — reducing inference time by 25% and cloud costs by 30% through model quantization and infrastructure optimization.

My RAG work spans Pinecone, FAISS, ChromaDB, Weaviate, and Milvus with hybrid semantic search at 92% retrieval accuracy, evaluated using Ragas and DeepEval. I've fine-tuned LLMs including Llama 3 using LoRA and QLoRA via Hugging Face, achieving 35% accuracy improvement over baseline. I integrate and orchestrate foundation models across AWS Bedrock, OpenAI GPT-4, Anthropic Claude, Google Gemini, and Mistral — selecting the right model per cost, latency, and task.

On the backend, I build scalable FastAPI microservices with async patterns, structured JSON/function-calling outputs, and security best practices including HIPAA-compliant data handling and JWT authentication. I prototype rapidly with Streamlit and Gradio before productionizing, and deploy with Docker, Kubernetes, and CI/CD pipelines with MLflow experiment tracking for zero-downtime releases.

I work independently, communicate clearly across technical and non-technical teams, and ship reliable AI products people use every day.