Senior Computer Vision Engineer

About the role

Join Swordhealth's Algorithms organization to advance the "Thrive" program, which provides AI-guided physical therapy for chronic pain. This role focuses on developing and deploying computer vision models that interpret human movement from camera feeds. You will be instrumental in turning real-time motion data into clinical insights for members.

Key facts

What you'll do

• Take ownership of core computer vision models, including 3D human pose and statistical body modeling, from initial concept to production deployment.
• Implement and optimize these models for real-time operation in both cloud environments and on-device on tablets.
• Manage the entire data and code lifecycle, encompassing training frameworks, annotation processes, pipelines, auto-labeling, test sets, and data taxonomy.
• Expand movement understanding by integrating multimodal approaches, combining vision with language and reasoning to create new solutions in movement intelligence.
• Standardize and improve model training, tracking, versioning, and deployment processes to ensure reproducibility and testability of all results.
• Design automated systems to scale the team's output beyond manual effort.
• Contribute to the growth of the Computer Vision team by establishing and promoting best practices and guiding principles.

Requirements

• At least 5 years of experience solving complex computer vision problems, with models successfully deployed in production.
• A strong foundation in software engineering, covering architecture, pipelines, MLOps, and the complete model lifecycle.
• Deep, practical expertise in modern computer vision techniques, including transformers and convolutional models, with specialized knowledge in detection, segmentation, tracking, pose, or 3D.
• A data-centric approach, demonstrated by creating custom datasets, building data flywheels, implementing active-learning loops, and treating datasets as source code.
• Solid understanding of multimodal AI, capable of developing systems that combine vision with language and reasoning as needed.
• Excellent written, asynchronous communication skills, with a habit of documenting thoughts and designs for others to build upon.
• Proven self-direction and full ownership, able to define plans from ambiguous problems and deliver results independently.
• Proficiency with PyTorch or JAX, the Python data stack, and comfort in learning new languages and codebases.
• Experience with AI-native workflows, embedding AI into daily processes, from agents to LLM-in-the-loop tools.

Nice to have

• Experience in pose estimation, body modeling, movement intelligence, or human-motion analysis.
• Experience working with large datasets and building full-lifecycle ML systems in a startup or scaleup environment, taking on diverse responsibilities.
• Experience automating experimentation or model-improvement cycles.
• A product-oriented mindset, empathy for users, and a desire to deliver significant impact to millions of members.

Skills & tools

• Computer Vision
• 3D Human Pose
• Statistical Body Modeling
• PyTorch
• JAX
• Python data stack
• MLOps
• Transformers
• Convolutional Models
• Detection
• Segmentation
• Tracking
• Multimodal AI
• Language Models
• Reasoning

Practical notes

Candidates must possess a valid EU visa and be based in Portugal; relocation assistance is not provided. Benefits include health, dental, and vision insurance, a meal allowance, equity shares, a remote work allowance, flexible working hours, work-from-home options, discretionary vacation, and office snacks and beverages. All candidates are assessed for AI fluency, with a minimum requirement of Level 1 (Explorer) and an expectation of Level 3 (Integrator) for this role.