Senior ML Engineer (Computer Vision)

Responsibilities:

• Own vision models end-to-end: data, training, evaluation, optimisation, and production rollout.
• Build and improve detection, segmentation, tracking, OCR/ALPR and face-recognition models for real customer use cases.
• Make inference fast and cheap: export to ONNX, optimise with TensorRT, and serve through Triton on GPU and edge hardware.
• Design evaluation sets and metrics that predict real-world accuracy, and catch regressions before release.
• Work with the Go platform team on how models are packaged, deployed, versioned and monitored.
• Explore vision-language models where they beat task-specific models, and ship them when they do.
• Set technical direction for ML, review designs and code, and mentor other engineers.

Requirements:

• 3+ years building machine learning systems, including computer vision models running in production.
• Expert in Python and PyTorch; strong with NumPy and OpenCV.
• Hands-on experience with modern detectors and trackers (e. g., YOLO-family, DETR-family, ByteTrack or similar).
• Production inference experience: ONNX, ONNX Runtime, TensorRT and/or Triton Inference Server, on GPUs.
• Experience with real-time video pipelines (RTSP, frame decoding, batching) and their latency and throughput limits.
• Comfortable with Docker, Linux and GPU environments; Kubernetes is a plus.
• Experience with vision-language or multimodal models is a plus.
• A track record of owning outcomes and leading technical work, not just running experiments.

Skills

Computer Vision, DETR, ONNX, OpenCV, PyTorch, Python, TensorRT, Triton, YOLO