PyData Meetup @ Source.ag

Important! If you're joining us from the beginning please come to the Source office (first floor Johan Huizingalaan 763A, 1066 VH Amsterdam). If you're coming later, the talks will be at The Kari (Johan Huizingalaan 400 1066 JS Amsterdam)

The PyData Amsterdam meetup is back once again in person! Source.ag has been so kind to host us for a round of interesting talks.

We'll hear from three great speakers about 'yield foreacasting at AI powered indoor farms', 'designing ML products that users will actually enjoy using' and on 'how to use data science and machine learning in trying to disentangle how genes contribute to a trait of interest in plants'

Schedule:
18:00 - Pizzas will be served at the Source office (first floor Johan Huizingalaan 763A, 1066 VH Amsterdam, Netherlands)
18:45: Move to the Kari (5 minutes walk), B.3 Johan Huizingalaan 400
18:50: Welcome and introduction by Cedric Canovas, director of Data Science at Source
19:00 - Estelle Rambier: Yield forecasting at AI powered indoor farms
19:40 - break
19:45 - Ayala Ohyon: Designing and building Miro’s clustering feature
20:25 - break
20:30 - Paul Boon: Using Machine Leaning to reveal the genetic architecture of vegetable traits
21:00 - networking

About Estelle:
I graduated 5 years ago with a master's in applied mathematics in Bordeaux (France). I joined Source last September, to work on AI powered indoor farm, tackling a safer, more reliable and climate resilient food production. This talk will focus on the journey of cracking the yield forecasting module, an essential step to optimize the cultivation strategies and enable autonomously growing. From my first week setting up API calls, to the final presentation to the end user, this is the story of a 5 months delivery. You will hear about how we leverage plant science literature in our models, the careful designs of our features accordingly, and how this model fits into our larger multicomponent AI solution.

About Ayala:
I work as a Data Science Lead at Miro, delivering intelligent features that enable effective and fun collaboration for Miro's whiteboard customers. After my studies Econometrics & Operational Research at the VU in Amsterdam, I followed a data science traineeship at Xomnia. At that time I was working at Schiphol Airport, where I was mainly concerned with passenger flows and forecasts. After that I joined Fourkind Consultancy for several years, where I tackled a wide range of topics: from optimizing Milka chocolate recipes to streamlining Mercedes car configurations. Now that I've joined Miro, I'm happy to contribute to turning Miro from a tool into a (fun and valuable) teammate!
In this talk I will walk through the Data Science but also the product and design work needed to create an intelligent feature that can delight users by making their mundane tasks more efficient. As appealing automating boring tasks with Machine Learning may sound, doing so blindly may actually displease the users.

About Paul:
I work as a data scientist at Enza Zaden, which I joined three years ago after an academic career in Cognitive Neuroscience. At Enza Zaden we aim to develop vegetable varieties that are more tasty, healthy, and climate resilient. Our vegetable seed production contributes to crop harvests daily feeding approximately half a billion people. Developing a new vegetable variety through breeding is a long process which often takes ten years or more. One of the goals of the data science team is to speed up this process by predicting important traits based on the underlying genetics. These traits can be extremely complex, involving hundreds of interacting genes. This talk will focus on how we use data science and machine learning in trying to disentangle how genes contribute to a trait of interest, and how we try to predict which plant crosses are most likely to result in the ideal combination of traits.