Nabeh, a subsidiary of Master Works, delivers AI-powered solutions that help organizations automate operations, boost efficiency, and drive innovation. Its expertise spans computer vision, LLM applications, speech-to-text, data intelligence, and low-code AI solutions.
Nabeh is looking for an AI Product Engineer (Computer Vision & VLM) to develop, integrate, optimize, and deploy Vision Language Models (VLMs), computer vision solutions, and real-time video analytics capabilities across its Baseer product portfolio. The role focuses on transforming live camera feeds, recorded videos, and images into actionable insights, structured events, and AI-driven workflows for enterprise applications.
Key Responsibilities
Computer Vision Development
• Develop and integrate computer vision models for object detection, tracking, classification, segmentation, and visual event recognition.
• Implement solutions using YOLO, OpenCV, PyTorch, and other modern computer vision frameworks.
• Design video analytics pipelines for multi-camera environments.
• Build use cases involving people counting, occupancy, object tracking, activity recognition, safety compliance, and operational monitoring.
• Implement video analytics workflows involving regions of interest, event triggers, and configurable business rules.
• Evaluate accuracy under real-world conditions, including variable lighting, occlusions, camera angles, and environmental changes.
Vision Language Models & Multimodal AI
• Integrate VLMs to analyze images, video frames, and visual events using natural-language instructions.
• Develop visual question answering, image understanding, scene description, and visual reasoning capabilities.
• Combine traditional computer vision models with VLMs to improve contextual understanding and event interpretation.
• Build video summarization, semantic video search, and natural-language querying capabilities.
• Develop multimodal workflows combining visual intelligence with LLMs, RAG, and AI agents.
• Evaluate VLM performance, accuracy, hallucinations, latency, and inference cost.
Real-Time Video Processing
• Design and implement video ingestion and processing pipelines using RTSP, GStreamer, FFmpeg, or equivalent technologies.
• Build scalable architectures for processing multiple concurrent video streams.
• Implement frame sampling, batching, inference scheduling, stream recovery, and efficient resource utilization.
• Optimize video processing performance to meet target FPS, latency, and infrastructure requirements.
• Transform video events into structured metadata, alerts, searchable information, and business insights.
• Work with both live camera feeds and recorded video sources.
AI Model Deployment & Optimization
• Deploy and optimize computer vision and VLM models on cloud, on-premises, and edge infrastructure.
• Work with NVIDIA GPUs and edge AI platforms such as NVIDIA Jetson.
• Support inference optimization using TensorRT, ONNX Runtime, or equivalent frameworks.
• Explore deployment and optimization on Qualcomm Cloud AI accelerators where supported.
• Implement model benchmarking, performance monitoring, and resource utilization tracking.
• Integrate models with video management systems and enterprise applications.
Product Engineering
• Develop backend APIs and services to integrate AI models into web-based products and dashboards.
• Build reusable computer vision pipelines and configurable analytics components.
• Collaborate with product managers and UI/UX engineers to deliver complete product features.
• Create configurable use cases rather than hardcoded customer-specific implementations.
• Contribute to product architecture, technical documentation, automated testing, and release processes.
• Own assigned features from development through deployment and ongoing optimization.
Requirements
REQUIRED TECHNICAL SKILLS/EXPERINCE
• 3–6 years of relevant experience.
• Strong Python programming experience.
• Hands-on experience with OpenCV, PyTorch, and computer vision models.
• Practical experience with object detection and tracking models, including YOLO.
• Experience processing images and video from cameras or recorded sources.
• Strong understanding of image preprocessing, inference pipelines, and model performance metrics.
• Familiarity with VLMs such as Qwen-VL, LLaVA, or equivalent.
• Experience developing backend APIs using FastAPI or similar frameworks.
• Familiarity with Linux, Docker, Git, and AI deployment workflows.
• Ability to troubleshoot model accuracy, performance, and video pipeline issues.
• Fluency in both Arabic and English, with strong verbal and written communication skills.
PREFERRED QUALIFICATIONS
• Experience with real-time multi-camera video analytics platforms.
• Knowledge of RTSP, GStreamer, FFmpeg, and video codecs such as H.264/H.265.
• Experience with tracking frameworks such as ByteTrack or DeepSORT.
• Experience with segmentation, pose estimation, OCR, and action recognition.
• Familiarity with video-language models and temporal video understanding.
• Experience with ONNX, TensorRT, model quantization, and inference acceleration.
• Experience deploying AI workloads on NVIDIA Jetson, NVIDIA GPUs, or Qualcomm AI platforms.
• Knowledge of video management systems, camera integration, and edge computing.
• Experience with Kubernetes and distributed inference architectures.
• Experience developing AI solutions for transportation, aviation, surveillance, industrial inspection, or smart cities.