Senior Computer Vision & Machine Learning Engineer

About ExpertVision Ai
ExpertVision AI is a Swiss computer vision startup building AI-powered loss prevention for physical retail. Our multi-camera platform detects theft in real time and includes patent-pending, non-biometric customer re-identification technology. We're already running pilots with large retailers and collaborating with top Swiss Universities.
We're a small, fast-moving team solving hard 3D perception and real-time systems problems in live store environments — not just research demos.
About the role
We are seeking a highly skilled and passionate Senior Computer Vision & Machine Learning Engineer to drive the development of advanced real-time perception systems, 3D spatial reconstruction, and video analytics pipeline solutions.
In this role, you will be responsible for building high-performance C++/Python video streams, deploying Deep Learning models optimized on NVIDIA GPUs, implementing 2D/3D human body reconstruction, and developing multi-camera object tracking systems operating over low-latency edge and cloud infrastructures.
Key Responsibilities
Real-Time Video Pipelines: Design, build, and optimize high-throughput, low-latency video streaming pipelines using GStreamer, NVIDIA DeepStream, and OpenCV interfacing with RTSP and ONVIF cameras.
3D Perception & Reconstruction: Implement advanced camera calibration techniques, single-view 3D reconstruction algorithms, and 2D/3D human body mesh/pose estimation pipelines.
Tracking & Analytics: Develop robust multi-object and human tracking systems capable of handling occlusions, dynamic lighting, and complex real-world edge cases.
Model Optimization & CUDA Acceleration: Optimize Deep Learning inference workflows (TensorRT/CUDA/C++) to run efficiently at high frame rates with low latency on GPU hardware.
High-Performance System Code: Write clean, highly performant C++ and Python code backed by solid profiling, concurrency, and memory management.
Event-Driven Architecture: Integrate vision telemetry and metadata into asynchronous messaging systems like NATS (or Kafka/RabbitMQ) for downstream processing.
Deployment & Linux Infrastructure: Containerize microservices using Docker and Docker Compose for seamless deployment across production Linux environments.
Technical Skills
Experience: 3–6+ years of professional engineering experience delivering production-grade Computer Vision and Machine Learning solutions.
Languages: Advanced proficiency in both Python and C/C++ (modern C++17/20, CUDA C++ integration).
Video Streaming & Acceleration: Hands-on mastery of GStreamer, NVIDIA DeepStream SDK, OpenCV, and NVIDIA CUDA acceleration.
3D & Geometry Vision: Deep expertise in camera calibration (intrinsic/extrinsic), epipolar geometry, single-view 3D reconstruction, and 2D/3D human body/pose estimation (e.g., SMPL, OpenPose, MediaPipe, mesh recovery).
Tracking Algorithms: Demonstrated experience implementing 2D/3D multi-target tracking algorithms (DeepSORT, ByteTrack, Kalman Filters, re-identification models).
Ingestion & Protocols: Experience working directly with IP/Security camera hardware, network protocols (RTSP, ONVIF), and hardware video decoders (NVDEC).
Performance Optimization: Demonstrated ability in high-performance computing, code profiling (NVIDIA Nsight, gprof), memory footprint reduction, and throughput optimization.
Messaging & Systems: Experience with message brokers (NATS, Kafka), containerized deployments (Docker, Docker Compose), and operational Linux environments.
Nice to Have
Familiarity with TensorRT custom plugin development in CUDA/C++.
Knowledge of edge orchestration or embedded platforms (NVIDIA Jetson AGX/Orin).
Background in synthetic data generation or domain adaptation for vision models.
Location
Remote — we're a remote-first team. Occasional travel to our Zurich HQ or team offsites may be expected. We want to hear from candidates based in:
Romania or
India
What we offer
Genuinely hard, unsolved problems in 3D perception and real-time tracking — running in live retail stores today, not academic benchmarks.
Architectural ownership: as a senior hire, you'll shape the core CV/ML pipeline and set technical direction for the team.
Direct collaboration with Swiss Universities.
Competitive compensation package for the right individual.

Work arrangement
Yes