Job Summary

The Computer Vision Engineer II – Vision Model Development & Automation will focus on building robust, production‑ready computer vision pipelines that solve real‑world problems in surveillance, industrial automation, and smart spaces. This role is designed for strong Tier‑1 fresh graduates (or 0–1 year experience) who combine deep knowledge of vision algorithms (both traditional and deep learning based) with the software engineering skills needed to automate, optimize, and integrate vision models into scalable platforms.

Key Responsibilities

• Develop, integrate, and optimize end‑to‑end computer vision pipelines: from image/video ingestion to pre‑processing, inference, post‑processing, and metadata generation.
• Implement and fine‑tune state‑of‑the‑art vision algorithms for tasks such as object detection, tracking, re‑identification, segmentation, anomaly detection, and OCR.
• Build and maintain automated training and evaluation pipelines (MLOps) that streamline model updates, benchmarking, and deployment across edge and cloud environments.
• Collaborate with platform engineers to optimize vision models for latency and throughput using tools like TensorRT, OpenVINO, DeepStream, or ONNX Runtime.
• Create robust logic for complex event processing: correlating multiple detections, applying temporal/spatial rules, and filtering noise/false positives to generate reliable business insights.
• Develop tools and scripts for dataset management: automated annotation, quality checks, synthetic data generation, and active learning loops.
• Troubleshoot vision performance issues in diverse real‑world conditions (lighting, occlusion, camera angles) and implement algorithmic fixes or data‑centric improvements.
• Document vision pipeline architecture, model interfaces, performance benchmarks, and deployment guides for internal and external consumers.

Skill Requirements

Must Have

• Strong programming skills in C++ and/or Python, with a focus on writing efficient, modular code for vision applications.
• solid understanding of Computer Vision fundamentals: image processing (filtering, morphology, geometry), feature extraction, and camera models/calibration.
• Proficiency with core CV libraries and frameworks: OpenCV, NumPy, and at least one DL framework (PyTorch/TensorFlow) applied to vision tasks.
• Hands‑on experience (projects/internships) developing and evaluating vision models: CNN architectures (ResNet, MobileNet, EfficientNet), detection families (YOLO, SSD, Faster R‑CNN), or tracking/segmentation approaches.
• Familiarity with video processing concepts: codecs, containers, streaming protocols (RTSP, HLS), and handling frame‑level data efficiently.
• Understanding of software engineering basics: version control (Git), unit testing, debugging, and integration of algorithms into larger applications.
• Strong analytical skills to evaluate vision performance: precision/recall/F1, mAP, IoU, MOTA, frame rate (FPS), and resource usage (CPU/GPU/RAM).

Other Requirements

• Experience with vision acceleration and deployment tools: NVIDIA DeepStream, TensorRT, CUDA, OpenVINO, GStreamer, or similar inference engines.
• Exposure to MLOps/DevOps for vision: CI/CD for models, automated testing of pipelines, Docker/containerization, and tools like DVC or MLflow.
• Knowledge of advanced vision topics: 3D vision (stereo, depth, point clouds), SLAM, multi‑camera tracking, or generative models for image synthesis/restoration.
• Familiarity with edge computing platforms (NVIDIA Jetson, Raspberry Pi, etc.) and optimizing models for constrained environments.
• Understanding of cloud services for vision (AWS Rekognition, Azure Computer Vision) or labeling platforms (CVAT, Labelbox, etc.).
• Contributions to vision‑related open source projects or participation in vision challenges/competitions.