Andrew has been coding for 40 years and has expertise in software development, product management, systems design and team leadership across a range of industries including science, technology, engineering, health, automotive, transport, mobile phone, and travel.
From 2009 - 2017 Andrew ran the Software development group at NIWA focussed on creating products and applications for climate, water, biodiversity and data science. I have run my own company and led a start-up mobile phone company software team through a high growth period. I have created and developed multiple agile cross functional teams, managed DevOps processes and modernised IT platforms including migration to cloud services.
Recent advances in machine learning and computer vision are enabling many new types of opportunities for scientific research and analysis services. Sometimes it seems as if any suitably annotated collection of images can yield observations and measurements about the real world that have previously required manual or expert observations or expensive sensors.
The CRI, Plant and Food Research, works in the areas of horticulture, viticulture and aquaculture and has developed diverse image recognition models that can for example: measure the growth of fish, detect diseases in grapevines and count kiwifruit pollen.
Image processing and ML prediction models take time to run - often seconds rather than milliseconds, so building a production scale system that can handle multiple models, large numbers of images and sessions in reasonable time requires an interesting mix of modern technologies.
Bypassing the core of how machine learning works, this workshop looks at the full stack of tech involved in constructing such a system. This includes: Docker containers, AWS deployments, Python, API and processing units, task queues and MongoDB.Register your interest