高级计算机视觉工程师/Senior Computer Vision Engineer

• Design & build a high‑throughput inference engine that can orchestrate multiple vision models (detection, tracking, recognition) on edge hardware.
• Accelerate model inference using TensorRT, ONNX Runtime, TVM, custom CUDA kernels, and hardware‑specific APIs (GPU, NPU, DSP).
• Create CI/CD deployment pipelines (containerization, model versioning, zero‑downtime roll‑outs) that move models from training to on‑cart runtime.
• Implement monitoring, profiling & alerting for latency, throughput, and resource use; iterate to meet strict real‑time SLAs.
• Integrate inference services with Caper’s broader infra (message buses, telemetry, OTA update system, configuration management).
• Collaborate with XFN teams (product, data, hardware, operations) to ensure the on‑board AI stack is robust, scalable, and can run A/B or store‑level experiments with minimal friction.
• Partner with research engineers to translate prototype CV models into production‑ready, inference‑efficient versions.
• Document architecture, standards & best‑practice guidelines; mentor junior engineers on infra‑focused development.
• Stay current on emerging edge‑AI frameworks, model compression, and multi‑model scheduling algorithms; evaluate and adopt them when beneficial.

• MS or Ph.D. in Computer Science, Electrical Engineering, or a related field (AI/Computer Vision focus a plus).
• 5+ years of professional experience building and maintaining AI inference infrastructure for edge or embedded systems.
• Strong software‑engineering background with expertise in C++ and Python, solid understanding of design patterns, CI/CD, and debugging.
• Proven experience with model deployment frameworks (TensorRT, ONNX Runtime, TVM, OpenVINO, Triton Inference Server, etc.) and hardware acceleration (CUDA, cuDNN, Vulkan, OpenCL, ASIC/NPU APIs).
• Hands‑on experience creating inference engines that orchestrate multiple models simultaneously while meeting real‑time latency budgets.
• Demonstrated ability to integrate AI components into larger distributed systems (REST/gRPC services, message queues, edge‑cloud orchestration).
• Excellent communication skills; comfortable presenting technical solutions to cross‑functional stakeholders.
• Excellent English documentation skills – must be able to write clear, comprehensive technical documents for a global audience.

• Familiarity with cloud platforms (AWS, GCP, Azure) and MLOps tools (Kubeflow, MLflow, SageMaker, etc.).
• A portfolio of computer‑vision projects (object detection, tracking, pose estimation) that showcases end‑to‑end pipelines from data collection to deployment.
• Knowledge of model training workflows, dataset management, and transfer learning – useful for close collaboration with research teams.
• Publications or open‑source contributions in efficient inference, model compression, or edge AI.
• Proficiency with GPU profiling tools (Nsight, PerfKit) and performance‑tuning techniques (kernel fusion, quantization, pruning).
• Good English communication skills – ability to present ideas clearly and collaborate effectively with globally distributed teams

Caper