I’m creating a lightweight Android app for medical-imaging students that turns chest X-ray files into plain-language reports with a single tap. The flow is intentionally stripped down—an “Upload Image” button, a “Generate Description” button, and a text area for the result. Behind the scenes the image is passed to an AI vision-language model such as Gemini or MedGemma (I’m happy to hear alternatives if you can justify accuracy and latency). Here’s what I need from you: • Build the Android front end in Kotlin or Java through Android Studio, keeping the layout to one activity or fragment with those two controls. • Wire that UI to the inference layer, whether the model runs on-device via TensorFlow Lite / ONNX or calls a cloud endpoint—you can pick the most practical route but it must stay free or low-cost for student use. • Return a concise description for each image and handle multiple selections gracefully (sequential processing is fine). • Provide an installable APK plus complete, well-commented source code and a short README explaining model setup, API keys (if any), and how to swap in a different model later. I’ll choose the developer whose past work best demonstrates Android builds that interface with ML or external APIs, so please include that in your bid.