Applied Scientist (Computer Vision)

About the role

You will architect and implement advanced motion detection capabilities that power real-time decision-making across both Edge and Cloud platforms. This role places you at the intersection of research and production, where your hypotheses will directly influence system behavior in diverse deployment environments. You will own the full lifecycle of computer vision experiments, from raw sensor or image data to validated model artifacts ready for integration. Your work will focus on extracting meaningful patterns from complex visual streams to ensure high performance and adaptability under varying conditions. You will translate abstract research goals into concrete algorithmic improvements that enhance detection accuracy and efficiency. Collaboration with cross-functional teams will be essential to align technical solutions with overarching product objectives. Ultimately, your contributions will define the reliability and scalability of motion detection features that serve both distributed Edge nodes and centralized Cloud infrastructure.

Key facts

What you'll do

Design and develop innovative machine learning algorithms specifically tailored for motion detection in dynamic visual environments.

Analyze large-scale video and sensor datasets to uncover hidden correlations and patterns that inform model architecture and training strategies.

Tune hyperparameters and optimize existing algorithms to squeeze out incremental gains in performance, accuracy, and inference speed.

Visualize complex research outcomes through clear, interpretable representations that facilitate stakeholder understanding and decision-making.

Evaluate emerging technologies in machine learning and assess their suitability for real-world production constraints and edge deployment scenarios.

Collaborate closely with colleagues from other departments to align technical approaches with business requirements and operational limitations.

Implement cutting-edge ideas from academic literature into robust, scalable production processes that can withstand real-world conditions.

Report progress and project milestones to management using precise metrics and clear narratives that highlight technical achievements and business impact.

Engage in continuous self-improvement by mastering new frameworks, libraries, and methodologies relevant to computer vision and motion analysis.

Mentor junior colleagues by sharing knowledge, providing code reviews, and elevating the overall technical competence and confidence of the team.

Validate model outputs through rigorous testing protocols to ensure reliability, consistency, and adherence to predefined success criteria.

Explore alternative sensor inputs and data sources to enrich motion detection pipelines and improve resilience in challenging scenarios.

Document experiments, decisions, and results meticulously to create a reproducible knowledge base for future research endeavors.

Champion best practices in model development, version control, and experiment tracking to maintain high standards of engineering excellence.

Requirements

Candidates must bring a proven track record of four or more years dedicated to machine learning projects with a primary focus on computer vision.

You must possess practical experience in at least one core problem domain such as classification, detection, or segmentation, demonstrating mastery of relevant techniques.

Proficiency in Python3 is mandatory, along with hands-on experience using NumPy, scikit-learn, pandas, and SciPy for data manipulation and analysis.

Deep learning framework expertise must include PyTorch, including model construction, training loops, and optimization strategies.

You should have direct experience deploying machine learning models into production environments, managing issues related to scalability and latency.

A solid grasp of machine learning and deep learning concepts is required, including loss functions, optimization techniques, and evaluation metrics.

Excellent written and spoken English skills are necessary to communicate effectively with distributed team members and stakeholders.

The role demands strict adherence to deadlines and the ability to work independently with minimal supervision while maintaining high-quality output.

You must be comfortable working within a fast-paced, iterative research environment where requirements evolve based on experimental findings.

Strong attention to detail is essential to ensure that experiments are well-designed, results are accurately interpreted, and models are thoroughly validated.

Commitment to continuous learning is expected, as the field of computer vision advances rapidly and new methodologies emerge frequently.

Collaboration skills are critical, as you will regularly interact with engineers, data scientists, and product managers to align technical and business goals.

Willingness to follow established coding standards and version control workflows is required to integrate smoothly with existing development pipelines.

An analytical mindset is necessary to diagnose issues, interpret experimental outcomes, and make data-driven decisions about model improvements.

Reliability and ownership are key, as your work will directly impact the performance of motion detection features used in production systems.

Nice to have

Practical experience with Generative Adversarial Networks (GANs) and Variational Autoencoders (VAEs) for advanced synthesis and detection tasks.

Familiarity with probabilistic programming and Bayesian frameworks to model uncertainty in motion and sensor data.

Knowledge of model optimization methods such as pruning, quantization, and knowledge distillation to improve efficiency and deployment feasibility.

Basic understanding of web and client-server architectures to facilitate seamless integration between Edge devices and Cloud services.

Experience with asynchronous programming using asyncio and aiohttp, along with other async libraries commonly found in backend systems.

Foundational understanding of Big Data concepts, including the distinction between MapReduce and in-memory processing paradigms.

Proficiency in core algorithms and data structures to support efficient implementation of complex computer vision pipelines.

Familiarity with SQL and NoSQL databases for managing experiment metadata, checkpoints, and large-scale training data.

Experience with containerization and orchestration tools such as Docker, Kubernetes, and Kubeflow to streamline deployment and scaling.

Programming skills in C++ and Bash scripting to support performance-critical components and automation tasks.

Practical notes

21 paid vacation days per year, paid public holidays according to the Ukrainian legislation.

Remote working mode is available within Ukraine only.

Free meals, fruits, and snacks when working in the office.

Medical insurance is provided from day one. Sick leaves and medical leaves are available.

Development opportunities like corporate courses, knowledge hubs, and free English classes as well as educational leaves.

Gig-contract.

Application instructions and deadlines are not provided in the source material.