Mirza Umer

AI engineer

Evren AI • Part-time

Sep 2025 - Present • 11 mos

Full Stack AI Engineer

Mechine Learning

Amazon • Part-time

May 2024 - Present • 2 yrs 3 mos

I work as a Full Stack Machine Learning Engineer building production-grade AI agents and automation systems for real client-facing products. My role covers the complete pipeline: designing the architecture, building the backend, integrating large language models, and deploying systems that businesses actually rely on every day. I specialize in Retrieval Augmented Generation, known as RAG, which lets AI agents answer questions using a company's own documents and data instead of generic responses. I have built RAG-based systems, including a legal consultant agent trained on real contract documentation, handling clause interpretation and retention queries behind a secure client login. I also build controlled response conversational AI, meaning chatbots and agents that stay within carefully defined boundaries, which matters most in sensitive fields like healthcare and legal services where an AI cannot be allowed to improvise. Alongside this, I have contributed to an AI-mediated negotiation platform that guides structured conversations between parties, and an AI-powered exam question generation pipeline that creates multimodal questions including diagrams and images. On the automation side, I build workflows using n8n and LangGraph that connect APIs, trigger actions, and remove repetitive manual work from a business entirely. This includes self-monitoring systems that automatically detect problems and draft fixes using an LLM before a human approves the final action, so nothing risky happens without oversight. My technical stack includes Python, FastAPI, LangChain, LangGraph, n8n, OpenAI and Claude APIs, PostgreSQL, and MongoDB. I have also worked on generative AI pipelines including custom LoRA model training for image-to-video genera