Computer Vision Engineer

We are seeking a Computer Vision Engineer with 3 to 4 years of experience to design, train, optimize, and deploy computer vision models into production environments. The role centers on image and video understanding, deep learning, and real-time inference systems. You will be responsible for the entire pipeline, from dataset preparation and model training to deployment and ongoing monitoring.

Key Responsibilities

Computer Vision:
- Build and train models for image classification, object detection, segmentation, object tracking, and OCR/visual understanding.
- Fine-tune pre-trained vision models to improve performance.
- Conduct error analysis and iterate on models based on data insights.

Data & Training:
- Prepare and preprocess image and video datasets, handling annotations, augmentations, class imbalance, and label quality.
- Design appropriate train/validation/test splits to ensure robust model evaluation.
- Track experiments and manage model versions systematically.

Deployment & Optimization:
- Deploy models as APIs, batch jobs, or real-time video pipelines.
- Optimize inference performance focusing on latency, throughput, GPU usage, and cost efficiency.
- Utilize tools such as ONNX, TensorRT, or TorchScript for model optimization.
- Monitor deployed models and retrain them as necessary to maintain accuracy and reliability.

Required Qualifications

- 3 to 4 years of professional experience in computer vision and deep learning.
- Strong proficiency in Python programming.
- Extensive experience with PyTorch and/or TensorFlow frameworks.
- Familiarity with OpenCV and torchvision or similar computer vision libraries.
- Solid understanding of convolutional neural networks (CNNs) and Vision Transformers.
- Hands-on experience with object detection and segmentation techniques.
- Knowledge of transfer learning methodologies.
- Ability to evaluate models using metrics such as Precision, Recall, Intersection over Union (IoU), and mean Average Precision (mAP).
- Proven experience deploying models into production environments.
- Working knowledge of Docker and Linux operating systems.
- Experience with GPU inference and optimization tools like ONNX and TensorRT.
- Ability to write clean, maintainable production-level code beyond exploratory notebooks.

Preferred Qualifications

- Experience with real-time video analytics applications.
- Familiarity with RTSP/CCTV camera streams.
- Knowledge of tracking algorithms such as ByteTrack or DeepSORT.
- Experience working with NVIDIA GPUs or Jetson embedded devices.
- Exposure to MLOps tools like MLflow, Weights & Biases, or DVC for experiment tracking and model management.

This role offers the opportunity to work on cutting-edge computer vision projects that impact real-world applications, with a focus on scalable and efficient deployment of AI models.