ArunKumar G

Staff Customer Engineer, Google

Arun Kumar is a Staff Customer Engineer at Google Cloud, where he works on Gemini Code Assist and AI-native developer experience. With a career spanning GitLab, Grafana Labs, and DevOps/SRE leadership, Arun has pioneered ways to embed AI into the software lifecycle. He loves turning complex platform challenges into catchy, AI-driven stories from AI Sherlock to AI exoskeletons. His current passion is showing how AI will obliterate paved paths and reinvent platform engineering.

The AI Native Android Developer: Inner Loop & Outer Loop with Gemini

Most AI for Android talks end at showing a snippet of Kotlin generated by a model. But if you’ve ever shipped an app, you know that’s just 10% of the job. In this session, I want to share how I’ve been using Gemini Code Assist inside Android Studio and the Gemini CLI to help with the tougher parts: wiring up Compose screens, creating test coverage on the fly, and even trimming down painful Gradle build times. I’ll also show how AI can be slotted into release pipelines so that it’s not just “helping with code” but actually supporting the whole lifecycle. This talk comes out of real experiments with enterprise teams where some things clicked instantly and others failed spectacularly. You’ll see live demos, a few patterns that genuinely saved me time, and some hard-learned lessons about where AI still struggles. Key Takeaways : 1. Why I think about Android work in two loops: inner (code/debug/test) and outer (build/release) and how AI fits into both. 2. Examples of using Gemini Code Assist for Compose screens and tests that actually made it into production. 3. How Gemini CLI plugged into CI/CD and helped cut down build headaches in my experiments. 4. Lessons from my failures - where AI didn’t work, and what I would do differently next time. 5. A realistic view of what’s possible now vs. what’s still hype in AI for Android.