- Strong hands-on experience with deep learning-based computer vision, including object detection, classification, tracking, and real-time video analytics

- Practical experience with CNN-based architectures such as YOLO (v5/v8) or similar, and ability to train, fine tune, and evaluate models using PyTorch or TensorFlow

- Experience building real-time vision pipelines for live video feeds (CCTV / streaming video) with low-latency constraints

- Solid understanding of video analytics concepts including frame sampling, motion analysis, temporal consistency, and object tracking across frames

- Strong understanding of image and video preprocessing pipelines including augmentation, normalization, and handling real-world data challenges such as low light, occlusion, motion blur, and varying camera angles

- Hands-on experience deploying CV models on edge devices such as NVIDIA Jetson, Raspberry Pi, or similar embedded platforms

- Exposure to model optimization techniques for edge deployment including quantization, pruning, or use of lightweight architectures

- Ability to design and own end-to-end CV pipelines, from data ingestion and annotation to inference, monitoring, and performance evaluation in production

- Experience working with Vision-Language Models (VLMs) or vision-enabled LLMs, and integrating vision model outputs with LLM pipelines for reasoning, event understanding, or summarization

- Experience collaborating with backend and DevOps teams for production deployment, including familiarity with Docker and basic MLOps practices

- Ability to evaluate and monitor model performance in production using appropriate computer vision metrics

Good to Have :

- Experience with edge inference frameworks such as ONNX, TensorRT, or OpenVINO

- Hands-on experience with video streaming and processing frameworks such as OpenCV, RTSP, GStreamer, or similar

- Exposure to multimodal AI systems combining vision with text (and optionally audio)

- Experience with multi-camera setups, camera calibration, or scene-level analytics

- Familiarity with LLM orchestration frameworks such as LangChain or LlamaIndex

- Understanding of edge AI security, privacy, and data compliance considerations in surveillance or industrial environments

- Experience working on real-world CV deployments in domains such as smart cities, retail analytics, industrial monitoring, safety systems, or large-scale surveillance.