Senior Computer Vision & Tracking Engineer

Description

We are looking for an exceptional Computer Vision Engineer, a true wizard in machine vision, object tracking, and intelligent image processing.

You can look at a complex engineering challenge, understand the underlying problem, design the right algorithmic solution, and turn it into a reliable, real-time product running on embedded hardware.

This is not just a model-training role. We are looking for a hands-on technical expert who understands the full pipeline, from image acquisition and sensor characteristics, through algorithm design and model training, to optimization, deployment, and validation on real hardware.

You will work with visible and thermal imaging systems, developing advanced vision capabilities for real-world applications where performance, latency, robustness, and reliability matter.

The Mission

Take vision algorithms from concept to real-time product. You will understand the underlying problem, design the right solution, and carry it through the full pipeline: image acquisition and sensor characteristics, algorithm design and model training, optimization, deployment, and validation on real hardware. You own the result, and you deliver detection and tracking capabilities that perform reliably in real-world conditions.

Why Join Skapion?

• Full-pipeline ownership: Work across the whole chain, from image acquisition and sensor characteristics through algorithm design, training, optimization, deployment, and validation on real hardware.
• Products, not prototypes: Turn advanced algorithms into reliable, real-time systems running on embedded hardware.
• Real-world stakes: Build vision capabilities for applications where performance, latency, robustness, and reliability matter.
• Visible and thermal imaging: Tackle low-light conditions, changing backgrounds, atmospheric effects, and sensor artifacts on real systems.
• Classical meets modern: Combine classical computer vision with deep learning, including state-of-the-art detection and tracking architectures.
• Solve the system, not just the model: Diagnose failures using data, algorithms, and system-level insight rather than parameter tuning alone.

Responsibilities

• Design and develop detection, classification, segmentation, and tracking algorithms for visible and thermal imaging systems.
• Own the full vision pipeline: image acquisition and sensor characteristics, algorithm design, model training, optimization, deployment, and validation on real hardware.
• Combine classical computer vision with deep learning, including state-of-the-art detection and tracking architectures.
• Build and refine single-object and multi-object tracking, including target association, track initiation, and track maintenance.
• Develop motion estimation, image stabilization, and camera-motion compensation capabilities.
• Optimize algorithms for real-time performance on embedded and edge platforms under CPU, GPU, memory, power, and thermal constraints.
• Diagnose failures using data, algorithms, and system-level insight rather than parameter tuning alone.
• Integrate algorithms with camera hardware, video pipelines, and embedded imaging systems.

Requirements

Computer Vision & Image Processing

• Strong fundamentals in computer vision, image processing, and visual geometry.
• Hands-on experience with object detection, classification, segmentation, feature extraction, and image registration.
• Deep understanding of tracking algorithms, including single-object tracking, multi-object tracking, target association, track initiation, and track maintenance.
• Experience with motion estimation, optical flow, feature tracking, image stabilization, and camera-motion compensation.
• Ability to design solutions that combine classical computer vision techniques with deep learning where appropriate.
• Strong mathematical foundations in linear algebra, probability, statistics, optimization, and estimation theory.

Machine Learning & Model Training

• Practical experience training, fine-tuning, evaluating, and deploying computer vision models.
• Strong understanding of deep learning architectures and frameworks such as PyTorch and/or TensorFlow.
• Experience with dataset curation, labeling, augmentation, class imbalance, model evaluation, and error analysis.
• Ability to diagnose model failures and improve performance using data, algorithms, and system-level insights, not just by changing model parameters.
• Familiarity with modern detection and tracking architectures, including transformer-based or other state-of-the-art approaches where relevant.

EO/IR & Imaging Systems

• Experience working with visible-light and/or thermal infrared imagery.
• Understanding of image quality, noise, contrast, dynamic range, frame rate, resolution, and sensor limitations.
• Ability to account for real-world imaging challenges, including low-light conditions, changing backgrounds, atmospheric effects, and sensor artifacts.
• Experience integrating algorithms with camera hardware, video pipelines, and embedded imaging systems.

Advantages

Real-Time & Embedded Implementation

• Strong programming skills in Python and C++.
• Experience with OpenCV and common scientific computing and ML libraries.
• Hands-on experience deploying computer vision algorithms on embedded or edge-computing platforms.
• Understanding of real-time processing pipelines, multithreading, memory management, data movement, and latency optimization.
• Experience profiling and optimizing algorithms under CPU, GPU, memory, power, and thermal constraints.
• Familiarity with FPGA acceleration, hardware/software partitioning, or FPGA-based image processing (a significant advantage).