Machine Learning Research Intern

🧑‍💻 Our Tech Team

They transform complex challenges into elegant solutions!

Their ambition is to have a positive impact on the everyday lives of hundreds of millions of consumers around the world by helping them shop smarter. Joko’s engineers work hand-in-hand with Joko’s product managers, from exploration, design, and roadmap prioritization to technical specification, implementation, and deployment in production. We have experienced firsthand that teams where engineers have a wide range of skills and where engineering and product collaborate closely ship the best features, offering a truly delightful user experience to our users.

Led by Alex, our CTO, they build an incredible experience across all parts of our technical stack, which has many ramifications!

🎯 What You Will Do

As a Machine Learning Research Intern at Joko, you will work on the automatic analysis of web pages, which is central to many core features of the smart web browser (and of the smart browser extensions) that we are currently building. For instance, your work will be key to delivering functionalities such as price tracking, product comparison, or universal one-click purchase on any e-commerce website, independently of specific website designs. Your work will also lay down the foundations of a radically new type of web browser, able to recognize and transform elements in real-time on any website, in order to offer a unified and smooth shopping experience to our users.

You will typically work on the following problems:

• Webpage classification: to detect if a webpage is a checkout page, a basket page, a product page for instance.

• Webpage element classification: typically to identify web page elements such as prices, product images, user reviews, product descriptions, product features, credit card fields, shipping details forms, etc.

• Automatization of user journeys: for instance, to automatically complete a checkout flow on behalf of the user.

Solving these problems requires digging deep into how webpages are transformed and rendered by web browsers, and finding the right level of abstraction for the algorithms. In particular, you will work on the DOM structure (Document Object Model), which is a tree data structure that plays a central role in the functioning of web browsers. As the DOM is a tree, it can typically be leveraged through graph neural networks. But as it also contains very rich text and image data, these graph algorithms can be combined with NLP and Computer Vision algorithms to achieve maximum performance. You will also work on real-time inference and the embedding of your algorithms on devices with limited CPU and memory.

The literature on these subjects is still in its infancy, and exploration will represent an important part of the internship, through experiments, literature reviews, and theoretical developments. You will have full ownership of your projects, and the liberty to orient the research direction of your internship based on your results and what you consider promising amongst the directions we determined.

You will work closely with the engineering team that will be here to help you integrate coding best practices into your research and to give you an insider look into modern software development. You will also have the possibility to integrate some of the algorithms you designed into our product and monitor their impacts on hundreds of thousands of users.

Your responsibilities:

• Research: You will work on all steps of the research process – you will formalize the objectives of your work, conduct literature reviews to have a deep understanding of the problems, design new algorithms, analyze them both theoretically and experimentally, and collect and transform relevant data for your experiments.

• Exploration & ownership: You will participate in orienting the internship towards research directions you deem valuable to our users.

• Implementation, deployment & monitoring in production: Helped by the engineering team, you will be responsible for integrating into our product the most scalable and robust algorithms you will have worked on. Finally, you will monitor their impact on our users.

• Processes: You will help improve our R&D tools, processes, and organization.