Description:
We are seeking an experienced Principal AI/ML Engineer to join an existing senior AI team working on complex, large-scale computer vision systems. In this role, you will provide deep technical expertise in pedestrian and vehicle detection, tracking, and counting, with a strong focus on improving model accuracy and reliability in dense, real-world environments. Working alongside a team of Senior AI Engineers, you will optimize and fine-tune YOLO-based models and end-to-end video analytics pipelines to achieve 90%+ counting accuracy. This is a highly practical, hands-on role where proven experience solving complex computer vision challenges in production is valued significantly more than certifications or formal credentials.
Responsibilities
- Lead the optimization of computer vision models for pedestrian and vehicle detection, tracking, classification and counting.
- Diagnose sources of model inaccuracy and implement improvements to achieve and maintain 90%+ accuracy.
- Develop, train, fine-tune, benchmark, and optimize YOLO-based object detection models and related computer vision architectures.
- Improve model performance under challenging real-world conditions, including extreme crowd density, occlusion, variable lighting, camera angles and high traffic volumes.
- Work with large volumes of video captured from deployed camera units and improve the end-to-end inference and counting pipeline.
- Evaluate model performance, establish appropriate accuracy metrics, and systematically identify false positives, false negatives, tracking errors and counting discrepancies.
- Provide principal-level technical guidance and collaborate closely with an existing team of Senior AI Engineers.
- Ensure models are scalable, reliable and suitable for deployment across mission-critical, large-scale environments.
Key Requirements
- 8+ years of engineering experience with 4+ years of hands-on computer vision experience.
- Principal or Lead-level ability to independently diagnose and solve complex AI/model-performance problems.
- Deep practical experience with YOLO and modern object detection architectures.
- Strong experience with object detection, multi-object tracking, pedestrian/vehicle counting and video analytics.
- Proven experience improving the accuracy of computer vision models deployed in real-world production environments.
- Experience handling challenges such as dense crowds, overlapping objects/occlusion, object re-identification, camera perspective and variable environmental conditions.
- Strong Python and modern AI/ML framework experience, ideally including PyTorch and OpenCV.
- Experience evaluating models using appropriate detection/tracking metrics and translating model improvements into measurable production accuracy.
- Ability to work effectively within an existing senior engineering team while providing principal-level technical leadership.
- Production experience and demonstrated technical capability are prioritized over certifications.