Computer Vision & ML Expert

Assess computer vision model quality

Evaluate AI systems that process images, video and other visual data. The opportunity is intended for specialists based in Tel Aviv and across Israel, including researchers, engineers and advanced graduate students with relevant technical depth.

Model and data evaluation

• Review image recognition, object detection, segmentation and visual reasoning outputs.
• Assess prediction accuracy, quality and robustness, identifying difficult cases, bias and failure modes.
• Create, review and improve annotations used in vision training pipelines.
• Give detailed feedback on model behavior, architecture choices and output quality.
• Write technical evaluations with concrete improvement suggestions.
• Work across natural images, medical imaging, autonomous systems and document understanding.

Assignments are completed independently on a flexible schedule. The work requires careful evaluation of both the data entering a model and the predictions it produces.

Required technical background

Strong foundations in computer vision are required, including areas such as classification, detection, semantic segmentation or pose estimation. You need practical experience with PyTorch, TensorFlow or JAX and familiarity with CNNs, Vision Transformers, diffusion models or related architectures.

You should be comfortable reading machine learning research and understand preprocessing, augmentation and annotation practices for visual data. Systematic review, clear technical writing and reliable independent work are important.

Preferred qualifications

• A master's or PhD in computer science, electrical engineering or a related vision or machine learning field.
• Publications in venues such as CVPR, ICCV, ECCV, NeurIPS or ICML.
• Experience with 3D vision, video understanding, generative models or multimodal systems.
• MLOps, experiment tracking or model evaluation pipelines.
• Medical imaging, remote sensing, robotics or autonomous driving experience.
• Large scale annotation and data quality workflows.
• Python and scientific tools such as NumPy, OpenCV and scikit-learn.

Remote contract terms

The hourly contract calls for ten to 40 hours weekly and provides flexible remote work. Tasks span multiple visual domains, with possible continued work or contract extensions as projects develop.