Senior Machine Learning Engineer (MLOps)
Paris | Remote /
✨ TL;DR ✨
💰 60k€-85k€* + Stock-Options/BSPCE
🇪🇺 We are flexible: you can work from anywhere in Europe and come twice a month in Paris (fully reimbursed), or come to the office more often.
🇬🇧 PhotoRoom is an international team and we work in English.
* We will adapt to the cost of living if it is higher than Paris.
✨ About us ✨
PhotoRoom provides photo editing software powerful enough to create outstanding images yet simple enough to be used without any training.
We leverage deep learning to translate pixels into objects, drastically simplifying non-creative tasks such as removing a background. Our mission: enable entrepreneurs and small businesses to compose images that stand out.
Founded by Matthieu Rouif (10 years of experience in mobile apps, built apps used by 100+ million users) and Eliot Andres (4 years of experience in large-scale computer vision), PhotoRoom was selected by Y Combinator in 2020. Talented engineers, designers, and marketers have joined the team since then, coming from Apple, Algolia, or Google among others.
Today, PhotoRoom is profitable and is used by millions of users every month. We improve our product and our technology with the support of world-class experts in the field like Yann LeCun, Roxanne Varza, or Holger Seim. PhotoRoom was featured by Apple as one of the most innovative photo apps in recent years.
As a small tight-knit team, our priorities are impact and iteration speed. We ship new versions every week and talk to our users daily. There is nothing we enjoy more than working side-by-side with another team member to improve our users' experience.
Become one of the first hires in a fast-growing start-up and make your mark on a product used by millions of people. Apply at PhotoRoom today ⬇
✨ Your core mission ✨
We're looking for someone eager to work with research engineers to ship models into production, modifying models to significantly reduce memory, computation while maintaining accuracy. Performance is key for our users: the faster and accurate our models are, the better their experience is.
As an ML OPs Engineer, you are the bridge between our research team and our users.
You work with our small research team to put models into production. This is a diverse and challenging role. A primary role is optimising models for production inference, i.e quantisation, compilation, pruning, conversion to mobile.
Secondarily, you will maintain and optimise production inference servers for both speed and cost.
Thirdly, you will bring strong software development ability to contribute to our research and server codebases. This includes maintaining our cloud platform deployment for training and our model logging and comparison codebase.
Computer vision solves real users' problems at PhotoRoom, it saves them time and helps them grow their e-commerce or small business.
Preferred experience in at least one of the following areas:
- Practical experience in Deep Learning, especially for computer vision tasks.
- Writing backend code for inference servers
- Strong software development skills
For successful candidates, nearly all of the following will be true:
- You have demonstrated abilities in industrializing research models.
- You have practical experience in optimising models for speed and memory for real time inference.
- You have significant experience in building applications that use machine learning component technologies.
- You are excellent at collaborating with a team to get work done.
- You are skilled at both building prototypes from scratch and writing maintainable code inside large existing codebases.
Requirements for this position include:
- Master's degree (or equivalent experience) in Computer Science or a related field
- Proficiency with modern deep learning frameworks
- 1+ years of experience with training, optimizing and deploying state-of-the-art deep learning models.
- Ability to write efficient, clean, and reusable code in Python and/or C++
- Strong communication and collaboration skills
- Ability and willingness to learn new technologies quickly