How Much Work Does it Take and What Is it Like to Integrate an Android SW Stack on a Gadget?

Accepted Session
Short Form
Scheduled: Tuesday, June 26, 2012 from 2:30 – 3:15pm in B204


We all know about the Android Open Source project and that in theory anyone can make an android device with their very own customised AOSP ROM. But, what is it like to work on something using AOSP. How deep is that rabbit hole anyway?


This will be a mostly non-technical talk about the scope of complexities and issues anyone doing an android based gadget (phone or otherwise) well likely bump into. There will be links a-plenty, but mostly it will be an overview that will help you approximate the scope and size of that android based project you may be contemplating. Hopefully, it will help you get a leg up on your own learning ramp up for your own projects.

This talk will consist of a number of topics I think may be interesting from my experiences ramping the Android activities at work for the Intel phones over the past 3 years, and a few side projects I’ve been playing with at home.

I will likely reference AOSP, Cyanogen-mod, arowboat and a few other SW projects. I’ll include HW targets such as; Intel’s Medfield phones, tegra based gtablet, and omap beagleboard.

Speaking experience

I've been working on Android at Intel since 2008 (and Linux kernel since 2001) and drove Intel's learning curve on Android at a time when it was more interested in other things. Now that there are multiple large teams doing android I'm still involved with a surprising amount of interesting issues and problems I wouldn't have expected.

Today I am the "android-linux-kernel-architect" which means I worry about how to update, harden, debug, integrate and migrate to new kernels in a cost effective way, including trying to upstream the kernel enabling of the Intel devices. I work in a large globally distributed team focussed on producing product ready images for Android devices.