Job Description
What You'll Do
* Maintain, optimize, and ensure the reliability of our complex, multi-stage computer vision pipeline, which includes models for semantic segmentation, object detection, image classification, and optical flow estimation.
* Set the technical vision and roadmap for specific parts of the CV pipeline, making key architectural and algorithmic decisions to drive performance and accuracy.
* Lead the adaptation of existing models and the development of new components for upcoming projects, translating business needs into technical solutions.
* Oversee the entire data lifecycle for your models, from defining clear annotation guidelines to managing data quality and iterating on the feedback loop with our in-house annotation team.
* Write clean, maintainable, and robust production-level Python code, adhering to strong software engineering best practices.
* Manage model training, experimentation, and evaluation on our on-premise Linux server infrastructure.
Qualifications
* Master's degree or PhD in Computer Science, or a related field.
* 5+ years of professional experience building and deploying computer vision models in a production environment.
* Deep, hands-on expertise with PyTorch.
* Proven experience across a range of computer vision tasks (e.g., object detection, semantic segmentation, image classification).
* Exceptional software engineering fundamentals with a focus on writing clean, testable, and efficient Python code.
* High proficiency in a Linux environment is mandatory. You should be comfortable and efficient using shell scripting, SSH, and command-line interfaces for daily work.
* Demonstrated experience working with data annotation pipelines, defining guidelines, and ensuring data quality.
* Excellent problem-solving skills and the ability to work independently and take ownership of projects.
* Willingness to embrace new ideas and feedback, work collaboratively with others, and continuously seek improvement.
Nice-to-Haves
* Experience working with large-scale datasets (TBs) and building efficient data processing pipelines.
* Experience with model optimization techniques (e.g., quantization, pruning, TensorRT/ONNX conversion).
* Experience mentoring junior engineers or providing technical leadership.
Our Tech Stack
Primary Framework: PyTorch
Infrastructure: In-house Linux servers, CLI, SSH
Annotation: In-house custom tooling
Additional Information
Contact
Igor Popov
At AUTO1 Group we live an open culture, believe in direct communication, and value diversity. We welcome every applicant; regardless of gender, ethnic origin, religion, age, sexual identity, disability, or any other non-merit factor.