[email protected]

I build production Computer Vision and Edge AI systems that turn real-world video into actionable insights, automation, and measurable operational improvements.

stanislavkuskov

kuskov_stanislav

With 9+ years of industry experience, I have worked on autonomous-driving perception systems, ADAS features, road perception, real-time video analytics, and Physical AI products.

I lead computer vision projects end-to-end from translating business requirements into ML tasks and designing data pipelines to model development, deployment, optimization, and production monitoring. My focus is not only on model accuracy, but also on reliability, scalability, inference cost, maintainability, and performance under real-world constraints.

I help companies:

• Automate visual inspection and reduce manual review
• Detect safety risks, anomalies, and infrastructure issues from video
• Move Computer Vision systems from PoC to reliable production
• Build scalable multi-camera and real-time video processing pipelines
• Deploy and optimize models for cloud and edge devices, including NVIDIA Jetson
• Improve dataset efficiency through active learning, self-supervised learning, and synthetic data
• Build and lead teams at the intersection of research, engineering, and product

Tech stack:

• AI & Computer Vision: PyTorch, OpenCV, Hugging Face Transformers, YOLO, VLMs
• Inference & Optimization: TensorRT, ONNX, Triton Inference Server, CUDA
• Video & Edge AI: NVIDIA Jetson, DeepStream, GStreamer, FFmpeg, ROS/ROS2
• MLOps & Infrastructure: Docker, MLflow, Airflow, S3/MinIO, SQL, Linux
• Languages: Python, C++

📄 Co-author of “Lidar Annotation Is All You Need” - arXiv: 2311.04777

Open to Lead Computer Vision / ML / Applied AI opportunities where I can help build technically ambitious AI products with clear real-world and business impact.

TECHNICAL SKILLS: