Computer Vision Engineer (Reindentefication

About the role

Everstar is building MilTech hardware and electronic solutions for active battlefield deployment. We are looking for a Computer Vision Engineer to join our R&D department and develop algorithms that will be integrated into thousands of products.

Key facts

What you'll do

• Implement and refine ML/DL algorithms for last mile targeting, focusing on small object detection and tracking.
• Develop production-ready code, optimizing algorithms for real-time performance on target hardware.
• Integrate DL models by building application logic for decision-making, frame sequence processing, and result validation.
• Monitor current research and SOTA approaches, prototyping and testing non-standard solutions to improve CV/ML quality.
• Provide technical leadership through architectural decision-making, code reviews, and mentoring middle-level engineers.

Requirements

• 5+ years of practical experience developing CV/ML algorithms for detection, segmentation, and tracking.
• Proficiency in C++ (C++14/17) including production code development, memory management, multithreading, and performance optimization.
• Deep expertise in classical CV combined with DL approaches.
• Experience with motion estimation and stabilization, including dense/sparse optical flow, ego-motion estimation, platform motion compensation, and frame registration.
• Knowledge of methods for detecting small or low-contrast objects.
• Proficiency in robust geometric methods: homography estimation, RANSAC, epipolar geometry, and image warping.
• Experience with classical tracking methods: correlation trackers, Kalman/particle filtering, data association (Hungarian, JPDA), and track-before-detect.
• Practical experience with PyTorch, OpenCV, and ncnn.
• Ability to act independently and take responsibility for technical outcomes.

Nice to have

• Product mindset: balancing research quality with implementation constraints and MVP development.
• Experience with model optimization: pruning, quantization (PTQ/QAT), knowledge distillation, and tensor/low-rank decomposition.
• Background in drone technology or robotics.
• Familiarity with Ultralytics and OpenVINO.

Skills & tools

• C++14/17
• PyTorch
• OpenCV
• ncnn
• DL/ML algorithms
• Geometric methods

Practical notes

• Benefits: 24 days of annual leave plus one day for your birthday, medical insurance after the trial period, and mental health support.
• Professional growth: Paid professional training, certifications, and coverage for attendance at conferences like CVPR, NeurIPS, and ICRA.
• Flexibility: Flexible schedule and up to 10 days of remote work per year.
• Compensation: Market-level salary with performance bonuses.
• Logistics: Support for military reservation (if documents are valid) and financial relocation assistance.
• Office: Workspace in Kyiv on the right bank.