Computer Vision Engineer in Better Medicine

We are using DL to improve patients’ lives. We build systems that help radiologists by accurately segmenting, measuring and characterizing cancerous lesions from Computed Tomography scans. While conducting clinical study on renal cancer, and having MPV segmentation models for liver and pancreas areas, our next strategic focus is lung cancerous nodules.

Required skills and experience:

• Strong experience (>5 years) developing deep learning and classical computer vision pipelines covering various tasks: detection, segmentation, classification, and registration.
• Solid knowledge of classical ML algorithms
• Linear Algebra and Statistics
• Writing high-quality, clean, maintainable, tested, and efficient code.
• Experience working with big-sized image datasets (>100 GBs)
• Upper-intermediate English, solid communication, and feedback skills.
• Highly desirable. Experience with 3D imaging data (ideally in the medical domain, CT, MRI, etc.). Practical experience with MLOps would also be relevant in the future (model deployment/monitoring/versioning, automated re-training, etc....)

Our tech stack:

• Deep Learning: PyTorch, PyTorch Lightning, Hydra, MONAI, DVC / MLflow (or other alternatives)
• Classical CV: OpenCV, ScikitImage, SciPy
• Classical ML: Scikit-learn
• Other: NumPy, Pandas, Matplotlib ( or any other alternatives)
• Infra: Google Cloud, Docker, CircleCI for CI/CD.

What the first six months in this role will look like:

• Increase sensitivity of the existing pipeline for detection/segmentation of lung cancerous nodules.
• Implement classification models for differentiating lung nodules into different texture types (solid, part-solid, non-solid), and measure clinically relevant features of the found objects (volume, surface area, long/short axis, etc.)
• -  Research scientific papers for gathering and re-implementing SOTA ideas on the given problematic
• Co-organizing the tight feedback loop between produced models and the domain experts (radiologists and other medical experts)
• Taking part in the clinical study in real hospital settings to prove the robustness of the developed algorithms.
• Collaborate with the product engineering team to integrate trained models into the existing radiological workflow (our in-house image viewer and annotator)
• Doing (and receiving) code reviews and providing (and receiving) feedback to/from your peers.
•  Development of the cancerous tumors dynamics estimation (3D image registration task) + setting up an active learning system for effective model re-training with the minimum radiologists’ input.

Benefits:

• You will get early-stage company stock options if you join us full-time.
• Co-authorship in research papers in top journals/conferences in the field
• Experience working in a product-driven healthcare company
• The freedom to solve problems your way.
• The possibility to have a real impact.
• A lot of development opportunities.
• We have flexible working time.
• We are a truly remote company. Our small team is currently dispersed across Tallinn, Tartu, London, and Lviv.

You are likely to succeed if you...

• Are a friendly team player

• Have a problem-solver attitude and ability to deliver iterable solutions in a

  collaborative and open way.

• Have the ability to work independently and come up with your own

  solutions to difficult challenges.

• Like taking charge of problems and having experience in leading

  dev/research efforts.

• Have great communication skills and the ability to articulate complex,

  technical concepts to non-technical audiences.

• High level of self-organization, ownership, and responsibility.

• Have both tech & product skills. Passionate about technology, products,

  science, and medicine.