Role Overview

Develop and improve computer vision models for sports video, including player and ball detection, tracking, event recognition, and identity association.
Build and improve camera calibration, homography, and field-registration solutions that map image coordinates into normalized field coordinates.
Analyze existing computer vision pipelines, establish baselines, identify weak links, and recommend practical improvements.
Improve tracking robustness across different stadiums, camera placements, broadcast styles, video qualities, and environmental conditions.
Design experiments covering data acquisition, dataset creation, augmentation, model training, fine-tuning, evaluation, and deployment readiness.
Analyze failure modes and implement improvements that increase accuracy, reliability, scalability, and robustness.
Adapt existing models and pipelines to support new sports, leagues, camera configurations, and video sources.
Partner with data teams on labeling workflows, dataset quality, validation processes, and human-in-the-loop improvement cycles.
Work closely with software, platform, and DevOps engineers to deploy computer vision models and pipelines into production environments.
Improve inference performance, scalability, monitoring, and operational reliability.
Establish evaluation metrics, testing processes, and quality controls to ensure model performance remains consistent over time.
Lead initiatives end-to-end, from early technical discovery and prototyping through production deployment and ongoing improvement.
Contribute to system design decisions that integrate computer vision, machine learning, backend services, operations, and client workflows.
Communicate technical tradeoffs clearly with internal teams, client stakeholders, and engineering leadership.

Requirements

Strong hands-on experience building and improving production-grade computer vision systems.
Proficiency with Python and modern machine learning frameworks such as PyTorch.
Experience with video-based computer vision problems, including object detection, multi-object tracking, event recognition, identity association, or video analytics.
Strong working knowledge of geometric computer vision, including camera calibration, homography estimation, projective geometry, and mapping image-space detections to real-world 2D or 3D coordinates.
Experience designing or improving tracking systems that handle occlusions, object interactions, identity preservation, noisy detections, and missing information.
Experience evaluating model performance, identifying failure modes, and implementing practical improvements.
Experience adapting models to challenging real-world data where video quality, camera angles, camera placement, and environmental conditions vary significantly.
Experience with transfer learning, domain adaptation, data augmentation, and fine-tuning models on domain-specific datasets.
Strong software engineering fundamentals and the ability to write clean, maintainable, production-quality code.
Ability to work independently, prioritize effectively, and drive technical initiatives to completion.
Strong communication skills and the ability to collaborate directly with clients and cross-functional engineering teams.

Tech Stack

Python
PyTorch

Benefits

Competitive compensation with benefits, paid vacation, and sick leave.
The opportunity to work with a globally diverse team of top engineering talent on the industry’s toughest engineering challenges.
Ultra-flexible working conditions – we provide a generous office equipment allowance so you can work from home, we can also provide you with a desk at an office/coworking facility near you, or use both. No business travel necessary.
An enjoyable, start-up work environment, with excellent opportunities for professional growth and development.
Flexible working hours – as a remote-first company, our focus has always been on getting the job done well, not when or where it gets done.

Janea Systems

Open Positions (1)

Engineering

Senior Applied Computer Vision Engineer