Senior Computer Vision Engineer

We are hiring a Senior Computer Vision Engineer to own real-world vision systems end to end: design, train, optimise and deploy, for real-time, multi-camera video analytics in security- and safety-critical environments, and to help shape the technical direction of our CV team.

This is a hands-on senior role with growing technical leadership. You will build and optimise detection, tracking and re-identification models and run them reliably in real time on NVIDIA hardware. We work remotely across the region, with occasional on-site collaboration in Singapore.

RESPONSIBILITIES

Computer Vision Model Development

• Design, train and improve deep-learning models for detection, tracking and re-identification (YOLO, DETR or similar).

• Build effective data, training and evaluation pipelines; improve performance through data, training strategy, architecture and optimisation.

• Make informed trade-offs between accuracy, latency, cost and real-world constraints.

Real-Time Multi-Camera Systems

• Build and improve real-time, multi-camera video pipelines, streaming ingest, detection, tracking and de-duplication.

• Work on cross-camera person / object re-identification, occlusion handling, and camera calibration / homography.

• Drive count and identity reliability toward compliance-grade levels.

NVIDIA Deployment & Optimisation

• Deploy and optimise models on NVIDIA GPUs and edge devices: TensorRT engine building, FP16/INT8 quantisation, batching across streams, latency / throughput budgeting.

• Build and maintain production inference services (DeepStream / GStreamer, APIs, containers) and keep them reliable under load.

Evaluation, Leadership & Collaboration

• Interpret metrics (mAP, IDF1, ID switches, etc.), analyse failure cases, and monitor models in production without ground-truth labels.

• Own significant CV problems from investigation to production; set engineering standards; mentor engineers; help define architecture.

• Translate product and business needs into practical solutions and communicate trade-offs clearly.

• Several years of production computer-vision / ML engineering, with demonstrated senior-level ownership of models running in production.

• Strong Python and PyTorch; object detection with YOLO / DETR; solid OpenCV and classical image processing.

• Hands-on NVIDIA GPU deployment with TensorRT, engine building and inference optimisation (FP16 / INT8), not only training models.

• Real-time video experience: streaming / RTSP and multi-stream or multi-camera pipelines.

• Multi-object tracking and/or person re-identification experience.

• Docker and Linux; cloud (AWS or GCP); Git-based workflows and CI/CD.

• Strong ownership, communication and mentoring; comfortable where requirements are not fully defined.

• Degree in Computer Science, Electrical / Computer Engineering or a related field, or equivalent practical experience.

PREFERRED QUALIFICATIONS

Any of the following is a strong advantage:

• NVIDIA DeepStream / GStreamer and Jetson or RTX-class edge deployment.

• Cross-camera re-identification and camera / multi-sensor calibration at scale.

• NVIDIA Triton Inference Server; synthetic-data generation or simulation (e.g. Isaac Sim).

• LLM / VLM / RAG and agentic frameworks for a video-intelligence layer.

• C++ or Rust for performance-critical components; MLOps tooling and infrastructure-as-code.

• A Master’s degree and/or peer-reviewed publications in CV / ML.

• Experience in a startup, scale-up or fast-moving product environment., We do not expect you to know every tool or framework. Strong fundamentals, sound judgment, honesty about gaps, and the ability to learn quickly matter more than checking every box.

If you enjoy solving hard, real-world computer-vision problems and want real ownership in a small team, we’d love to hear from you.