Machine learning engineer in video understanding

Scandit; United States; Hybrid; USD 150,000 - 190,000; Full time

Benefits offered by this job

Flexible hybrid or home working

Birthday off

Equity plan

Hackathons

Job summary

Scandit ищет ведущего инженера по компьютерному зрению и машинному обучению для разработки и внедрения end-to-end CV/ML систем, включая обработку видеоданных и развертывание моделей в продакшене.

Кандидат должен иметь опыт работы с трансформерами для видео и сильные навыки Python/ML-фреймворков, а также уметь сотрудничать с межфункциональными командами и общаться на английском. Гибридный формат работы в США.

Qualifications

• Должны быть практические навыки компьютерного зрения и машинного обучения; опыт разработки CV/ML систем от начала до развертывания
• Опыт работы с видеоп pipelines и распознаванием действий
• Умение работать в быстроменяющейся среде и ясно общаться на английском

Responsibilities

• Определять и приоритизировать технические эксперименты для валидации рисков CV/ML.
• Проектировать и внедрять конвейеры CV/ML на реальных видеоданных.
• Настраивать workflow аннотирования для подготовки датасетов.
• Сотрудничать с PM и руководителями инженерии для превращения бизнес-требований в тех roadmap.
• Подбирать архитектуры CV для распознавания действий в неупорядоченных условиях.
• Обеспечивать техническое лидерство в области видео CV и помогать формировать практики и найм.

Skills

Computer vision

Machine learning

Python

English communication

Cross-functional collaboration

Experiment prioritization

Production ML systems

Education

Bachelor's degree in CS/EE/Math

Tools

Python

PyTorch

TensorFlow

Vision Transformers

Job description

Описание:

Scandit gives people superpowers. Whether enabling delivery drivers to make quicker deliveries, matching a patient with their medication, or allowing retailers to make store operations more efficient, our technology automates workflows. It provides actionable insights to help businesses in a variety of industries. Join us as we continue to expand, grow, innovate, and help take Scandit to the next level.

Задачи:

• Defining and prioritizing technical experiments to validate key risks (e.g. precision, recall, latency) for computer vision models in retail stores
• Designing and implementing end-to-end CV/ML pipelines on real-world video data, from data ingestion and preprocessing through model training, evaluation, and deployment
• Setting up and refining the human annotation workflow to create high-quality training and test datasets from video streams
• Collaborating with product managers and engineering leaders to clarify requirements, align on milestones, and translate business needs into technical roadmaps
• Selecting, adapting, and optimizing state-of-the-art computer vision architectures for suspicious behavior and action detection in unconstrained environments
• Providing technical leadership in the space of video-based computer vision applications, helping shape best practices and future hiring for the team

Требования:

• University degree in Computer Science, Electrical Engineering, Mathematics or a related technical field (advanced degree is a plus but not mandatory)
• Significant hands-on experience in computer vision and machine learning, including designing, training, evaluating and deploying models in production
• Proven track record of building end-to-end CV/ML systems from 0→1 (ideally commercial products rather than purely academic projects)
• Strong expertise with video-based pipelines and human action or behavior recognition (e.g. retail, autonomous checkout or similar domains).
• Hands-on experience with vision transformers for temporal video understanding.
• Solid software engineering skills (e.g. Python and modern ML/CV frameworks) and experience working with real-world, noisy data
• Ability to work in a dynamic, high-uncertainty environment, prioritize experiments, and iterate quickly on technical approaches
• Ability to collaborate effectively with cross-functional teams (e.g. product, backend, data, operations) and communicate complex technical topics clearly in English (B2 level or higher)
• Eligible to work in the hiring location

Будет плюсом:

• Experience fine-tuning Vision-Language Models (VLMs) is a strong plus;
• Experience with privacy-aware or GDPR-compliant data pipelines, especially for video data;
• Background in building or selecting annotation tools and workflows for computer vision training data;
• Prior work on retail, autonomous stores using computer vision

Условия:

• A highly skilled team and a fun environment where you can put your enthusiasm for cutting-edge technologies to use
• Hackathons
• Flexible, office, hybrid or home working
• People-first culture
• Global team outings
• Festive/end of year all company celebrations
• Your birthday off
• An attractive individual equity plan in a high growth company
• Excellent office infrastructure, optimized for hybrid working in Zurich, Warsaw, Tampere, and London.
• Excellent support for remote work across Switzerland, Finland, Poland, UK, Italy and Germany
• Specific benefits related to the location you are joining