Data Scientist

Contenu de l'offre Data Scientist chez Akur8

Paris (75)

Akur8 develops a Saas platform which helps insurance companies in their pricing strategy to ensure that their prices as are competitive as they can be. To do so, we rely on cutting-edge AI modelling and heavy cloud-computing, leveraging the most recent technology.

You may know pretty well how to build complex models using state of the art machine learning techniques. However, when it comes to describe a complex phenomenon with a simple and interpretable model, you may have no other ideas than to ‘hack’ a Gradient Boosting or a Multi Layered Perceptron. With researchers from ENS-INRIA we built the fundamentals of an optimization technique that is interpretable by design. We bring non-linear penalized regressions to the next level using state of the art optimization advances: if you're curious about the subject, you can take a look at our collaboration paper with a Google AI researcher.

As a data scientist, you will contribute to the design of machine-learning models to enhance risk modelling, demand modelling (probability of potential clients to buy our products or renew their contracts) and price optimization (determine best price to sell the insurance product).

Practically you will begin by... :

Reading, understanding and explaining to the team strengths and weaknesses of research papers in ML, algorithmic and optimization Implement pragmatic solutions and test their robustness on our data Optimize your code using Cython, to achieve the same performance of open-source packages such as Numpy or Scikit-learn.

... and integrate your findings in production:

Implement reliable tests, use continuous integration and exploit at best AWS Cloud capabilities with our back-end engineers.

Your work will:

Rely on a strong understanding of the machine learning methods and optimization techniques used (R&D papers) Take into account the business objectives of the models created (in particular: robustness, interpretability and implementation cost). Design, build, integrate and maintain efficient, reusable and reliable code.

Requirements

High proficiency in mathematics (in particular statistics), algorithmic (optimization), and programming. Knowledge in machine learning, signal processing, speech recognition or image processing. Interest in R&D topics Strong problem solving skills: high level of rigor, integrity, curiosity and self-motivation. Communication and presentation skills. Excited about code craftsmanship, to build robust code with the best practices. Comfortable with Python (NumPy, Pandas, ...) and knowledge of an object-oriented language (Java, C#...)

Bonus:

Experience with Scikit-Learn API and its coding standards. Experience with Docker, Terraform is a plus.

Benefits

You'll join a motivate team, willing to share their knowledge and passion for advanced computer science and software development.