Building and shipping working data science products is hard - learn from 10 years of Ian's experience to find efficient ways through the mess of bad data, difficult team communication and poorly maintained code through to successfully deployed projects. The talk will include ways of getting data, cleaning and debugging it, approaches to deployment and various tips I've picked up along the way that'll save you time. If you frequently start new projects or you're involved in deploying working systems then this talk should have something for you.