Senior/Lead Deep Learning Engineer
Time zone
Europe/USCommitment
Full-time, remotePersonal recruiter
Anastasiia ZhukDirection
Deep LearningCategory
Product Vacancies for products developed by TeamDevProject
Pure PhotosAbout the project
We invite an experienced Deep learning Engineer to join the team working on our commercial product — Pure Photos. It’s the service for analyzing and processing a large number of photos.
The product is already one of the bests in the field and strives to dominate. We process about 300 thousand images per month, and this is only the beginning. By developing AI, we solve and automate tasks that previously could only be performed with human participation. The product helps professional photographers automatically process photos. The main function is to remove background from photos. And we have more ideas to expand the functionality, for example, retouching and color correction. You can try it for free on the site and see the top quality provided for yourself.
300,000+
processed images a month
15+
releases a year
260+
merged PRs a year
3,400+
commits a year
280+
types in API
The team
4
Deep Learning Engineers
3
Web Ninja
Processes and instruments
Pure Photos team has built an elaborate system, which is based on the state of the art deep learning and computer vision algorithms. It provides an AI-based solution for recognizing images of people located on the most complicated outdoors backgrounds, and extracting them with surgical precision.
The solution was conceived and built from the ground up using Google Cloud Platform Services. We’re utilizing Google Compute Engine for our deep learning training as well as inference using the latest GPUs. Moreover, the Pure Photos team has built a custom solution that allows scaling up instances in different regions and zones with the ability to utilize preemptible VMs, which reduces the costs dramatically.
The application is built with the event-based approach built on top of the Spine Event Engine CQRS/ES framework. Pure Photos has all the business logic hosted in various AppEngine microservices as well as the front-facing web application. This allows us to handle almost any load with automatic upscale. Minimizes costs when the number of clients is low with the ability to downscale to 0.
We use modern CI/CD tools such as Terraform and Packer that allow us to implement an “Infrastructure as a code” approach and automate deployments.
- Java, JavaScript, TypeScript, Python, NodeJs, and Vue.
- JUnit, Mocha, Karma.
- DDD, CQRS, Event Sourcing, Spine Framework.
- IntelliJ IDEA, WebStorm, PyCharm.
- Git, GitHub, GitHub Actions.
- Google Cloud Platform, Firebase platform, Terraform, Packer, Docker.
Our principles
In our daily work and when making key decisions, we are guided by the following principles:
- Confidence. We trust each other and our clients trust us.
- Support. We support each other in order to achieve our goals and boost professional growth as a team and as an individual.
- Quality. We pay attention to what and how we develop. We want to be satisfied with the results of our work.
- Flexibility. We are ready to adapt to changes if needed.
- Fun. We strive to enjoy what we do.
What you will do on this project?
Create and train deep learning models for computer vision: image classification, semantic segmentation, computational photography.
Knowledge and skills required
- 5+ years of experience in Engineering.
- 3+ years of experience in Deep Learning Engineering.
- Deep knowledge in math (linear algebra, calculus), Deep Learning, and Computer Vision.
- Experience with Python.
- Knowledge of PyTorch, NumPy, OpenCV, TensorFlow.
- Experience with Linux.
Would be a plus
- Experience in using NLP.
- Experience with Google Cloud Engine.
