Computer vision has been applied successfully to many fields in recent years, but not yet weather forecasting. That is about to change as our computer-vision-based tool using satellite images and data can bring hurricane and typhoon intensity prediction to a new level. We invite you to join us on an exciting pioneering journey in building something brand new with significant impact on the real world.
Machine Learning Engineer
Full-time (part-time possibility for students) in Helsinki, Finland
We are looking for developers to join us on a journey to the next front of machine learning: weather forecasting. We have openings for both senior engineers with the possibility to take tech management responsibilities as well as for junior engineers.
What is Hurricane Unwinder?
Computer vision has been applied successfully to many fields in recent years, but not yet weather forecasting. That is about to change as our computer-vision-based tool using satellite images and data can bring hurricane and typhoon intensity prediction to a new level. We invite you to join us on an exciting pioneering journey in building something brand new with significant impact on the real world.
Hurricanes, typhoons and tropical cyclones cause damages in the tens of billions of dollars and thousands of deaths annually. A one-point change on the five-point intensity scale translates on average to seven-folddifference in damages, For example: a category 4 storm causes 250 times more damages than a category 1. Yet it is one of the most poorly forecasted weather parameters at the moment. Better intensity information leads to more accurate warnings, more efficient physical and financial disaster management and, ultimately, together with hardware development, to successful weather modification (yes, weather modification, meaning downgrading of dangerous storms). Key customer segments for the forecast are governments for disaster management and holders of collateralized reinsurance, both of which are multi-billion-euro industries.
Weather is more complex than, say, facial recognition. Backed up by years of academic research, we now know how to combine pattern recognition of important features seen in hurricane satellite images with all the traditional data. We have won the Copernicus Masters innovation competition in the Disaster Management category and our product has international demand, which due to climate change advancing at an accelerating pace.
Team and role
Hurricane Unwinder was incorporated in March 2019 and we are backed by a supportive VC fund icebreaker.vc. Our team consists of Svante Henriksson with 11 years’ experience of climate research at Finnish Meteorological Institute and the South African Weather Service (https://fi.linkedin.com/in/svante-henriksson-5a5877113), seasoned entrepreneur Antti Pasila who has grown several companies (e.g. Kiosked, Claned) (https://www.linkedin.com/in/anttipasila) as well as three great ML engineers and world-class scientific and business advisors.
Your main responsibility will be in backend development involving deep learning. Our goal is to build a MVP product by October, the end part of hurricane and typhoon season 2019. We have agreed on one pilot project involving the Vietnamese government and several others are under discussion. The backend of our service runs on AWS SageMaker. It combines all the different data streams from visual satellite images to sea surface temperature measurements to open-source weather forecasts in as accurate a forecast as possible. Some of the tasks identified in addition to developing the algorithm overall are managing the pipeline of huge weather and satellite data from different sources all the way until the final data stream or output on our graphical user interface. Our current user interface is a webapp built with JavaScript & React. UX design is by necessity a key activity of the company as current interfaces on the market are unsatisfactory (https://www.technologyreview.com/s/612302/designers-are-reinventing-hurricane-maps-for-an-era-of-extreme-weather/). However, the UI is handled by other team members and you don’t need to be an expert on this part of the tech stack.
Experience & Skills
We hope that you have experience or studies in
Python
Deep learning, CNN/RNN
PyTorch or Tensorflow
Cloud computing, AWS
An interest in the topic and the natural sciences in general is necessary for you to have the right motivation, but no pre-skills are necessary as these can be learnt from the team. Conversely, if you have a strong background in the natural sciences, we can train you in machine learning if you have a strong basic foundation in computer programming.
Compensation & benefits
We offer a flexible and pleasant work environment and a pay increasing with the performance of the company as well as a generous amount of shares in the form of stock options. The exact amounts depend on your level of expertise and are negotiable. Remote work can be negotiated.
You’re more than welcome to get in touch for further information: svante@hurricanelab.io.
This job comes with several perks and benefits
Time is precious. Make it count. Morning person or night owl, this job is for you.
Get your caffeine fix to get you started and keep you going.
Easy access and treehugger friendly workplace.
Social gatherings and games; hang out with your colleagues.
Want to be a partner? Look no further.
We want to take you from good to great.