Middle ML Engineer

Location: Lviv (On-site)

Type: Full-time
Language Requirements: English (Ukrainian)

About the Project

We are developing next-generation autonomy systems used by the Armed Forces of Ukraine. Your work will directly impact national defense capabilities and reshape the future of autonomous aerial warfare.

The drone autonomy systems we are building turn math, algorithms, and simulation insights into guidance and decision-making logic for autonomous flight. The problems are ambiguous, physically constrained, and technically demanding.

Responsibilities

• Build ML-based computer vision models for onboard drone perception.
• Use classical CV methods when they solve the problem better than neural networks.
• Work with real and simulated visual data: collect, clean, label, inspect, and improve it.
• Train, evaluate, and debug models using clear metrics and failure analysis.
• Improve robustness under motion blur, vibration, low light, occlusion, and noisy sensor data.
• Optimize perception pipelines for onboard inference with limited compute, memory, and power.
• Turn experiments into reliable code that can be tested, integrated, and deployed.
• Communicate assumptions, tradeoffs, and findings clearly to the rest of the team.

Required Qualifications

• Practical experience with ML-based computer vision.
• Deep understanding of how neural networks work, how they are trained, and why they fail.
• Strong Python skills and experience with PyTorch, OpenCV, NumPy, and common ML/CV tooling.
• Strong understanding of training loops, loss functions, optimization, regularization, validation, metrics, datasets, and error analysis.
• Familiarity with classical CV: image processing, camera geometry, calibration, feature matching, optical flow, or tracking.
• Ability to debug model behavior using data, visualizations, logs, and experiments.
• Comfort with ambiguous problems where the right answer is not obvious at the start.
• Careful analysis, clear communication, and willingness to challenge weak assumptions.
• High ownership: the ability to take a hard problem, structure it, test ideas, and keep moving.
• Advanced agentic coding skills: the ability to use AI coding agents effectively while maintaining technical judgment and code quality.

Preferred Qualifications

• Experience with autonomous systems, robotics, drones, or other physical systems.
• Experience with visual perception for navigation, guidance, obstacle avoidance, or mapping.
• Experience optimizing inference on edge devices using TensorRT, ONNX Runtime, quantization, or similar tools.
• Experience with dataset pipelines, annotation workflows, synthetic data, or data augmentation.
• Experience using C++ for performance-sensitive CV or inference code.

What We Offer

• The opportunity to shape a mission-critical defense technology from the ground up
• A fast-paced result-oriented culture where decisions are driven by data and results
• Competitive compensation
• Access to state-of-the-art testing facilities, hardware, and simulation platforms
• Direct impact on Ukraine’s national security and technological edge