In this project, I build an AI-powered medical imaging agent that can look at X-rays, MRIs, CT scans, and ultrasound images using an agentic workflow powered by Gemini Flash. The whole point is to move beyond “just call an LLM” and actually design an agent that can take an input scan, run the right analysis step-by-step, and produce a clean, structured output a doctor (or a medical workflow) can consume. I also connect this to the agentic design patterns I’ve been teaching—because without the right pattern, these apps turn into brittle demos fast. I walk you through what the agent is doing, why multimodal + agentic orchestration matters in healthcare, and how you can demo the system end-to-end. You’ll see how the agent can surface potential issues that might be missed in a quick manual pass, while still keeping the mindset that this is an assistive tool—not a replacement for clinical judgment. If you’re serious about GenAI system design, this is the kind of “real project” that forces you to think about inputs, tools, guardrails, and outputs like an engineer, not just a prompt-writer.