How to pick an AI project that will succeed

After mentoring 167 AI projects for Data Science Retreat, Jose Quesada knows better than most what it takes to pick an AI project that succeeds. He have distilled this knowledge into a set of rules to pick good AI projects.

• Talk -

In this talk Jose will cover:

• How you don't need all that much data (Minimum viable corpus), thanks to pretrained models
• How your idea should pass 'eyebrow test.'
• How you only need to be in the ballpark of a good idea
• How picking a problem that moves you helps to avoid pitfalls
• How to produce unique data using hardware (sensors, cheap electronics)
• How you need a gold standard
• How computing at the edge avoid privacy problems
• How to avoid projects with 'the gorilla problem.'

• Bio -

Jose is the founder and CEO at Data Science Retreat (DSR), the most advanced machine learning course in Europe. He specializes in practical applications of machine learning that ship in less than three months. After managing 167 machine learning projects (Computer vision, natural language processing, recommenders), Jose believes there's lots of low hanging fruit in machine learning, and it's tragic most companies are not doing much.

His company has trained staff for Audi, Zalando, GfK, Hella Aglaia, and Uber. He has also created training materials for IBM through Lightbend (advanced machine learning for fast data/Spark).

Jose holds a Ph.D. and has been doing ML for 20 years now, and has started businesses in Berlin, San Francisco, and Toronto.