Lead Data Scientist - Computer Vision
- - - - - - - - - - - -
- End-to-end ownership & development deep learning based computer vision models.
- Understand and follow IT & company processes for machine learning projects.
- Understand the problem, analyze & develop approach to solve it
- Collect, analyze, review, clean, organize & document image datasets efficiently / programmatically
- Responsible for getting dataset annotated with help of annotators (internal/external)
- Transform images using opencv library, develop scripts to process annotation data/files
- Customize standard deep learning based computer vision models for object/keypoint detection/segmentation (using python & keras/tensorflow/pytorch)
- Train models systematically with clear rationale behind approach used
- Tune hyperparameters and loss functions
- Track training experiments and evaluate & compare models objectively
- Leverage other machine learning approaches like clustering, regression, classification to solve sub-problems
- Review, refactor, optimize for speed/efficiency and check-in code frequently
- Capture detailed documentation of data, models, application architecture, steps to execute code.
- Manage datasets and deploy models & inferencing pipelines on Azure cloud for scaling in production
- Communicate results/status updates to stakeholders once a week