Junior Data Scientist (Trainee)

About the Role

We are seeking a Junior Data Scientist to engage in the training, assessment, and refinement of computer vision models specifically tailored to the defence technology industry. This role offers the chance to work with complex, real-world datasets.

Key Responsibilities

• Clean and prepare drone image segmentation datasets for use.
• Support the processes of model training and performance evaluation.
• Extract valuable features from data of varying quality.
• Maintain and enhance pipelines for data generation and consumption.
• Create both manual and automated high-quality annotations.
• Focus on prioritizing data that delivers the most significant impact for model improvement.

Qualifications and Skills

• Strong foundation in Python programming, including numpy, OpenCV, and handling JSON/CLI scripts.
• Familiarity with fundamental PyTorch operations such as model training and fine-tuning.
• Understanding of different segmentation techniques—detection, instance, and semantic—as well as experience with COCO and YOLO data formats.
• Knowledge of evaluation metrics like mAP and IoU, and the ability to interpret training progress through curves.
• Comfort working with unorganized or noisy data along with familiarity with Git version control and Linux environment.

Additional Preferred Experience

• Experience using annotation tools like CVAT or SAM and working with Ultralytics YOLO or DETR/Hugging Face transformers.
• Insight into data augmentation techniques and fine-tuning strategies, including utilizing cloud-based GPUs such as AWS.
• A strong curiosity and enthusiasm for working with data.
• An appreciation for data quality and an understanding that poor labels result in poor model training outcomes.