Chief Computer Vision And Deep Learning Engineer Job

Key responsibilities:

a. Dataset Collection, Preparation, and Preprocessing
1. Develop strategies for data collection from diverse sources, including public datasets and real-world captures.
2. Oversee data cleaning, augmentation, annotation, and preprocessing processes, ensuring high-quality datasets for model training.

b. Model Training and Fine-Tuning
1. Lead the selection and design of appropriate deep learning architectures for object detection, segmentation, and tracking.
2. Oversee training experiments, hyperparameter tuning, and fine-tuning of models using frameworks like PyTorch and TensorFlow.
3. Develop and implement evaluation metrics and cross-validation techniques to ensure model robustness and accuracy.

c. Model Deployment and Optimization
1. Apply advanced optimization techniques (quantization, pruning, knowledge distillation) to enhance model efficiency for deployment on resource-constrained hardware.
2. Supervise the deployment of models on single-board computers (Jetsons, RPIs) and ensure real-time performance.

d. Model Explainability
1. Implement and oversee the use of interpretability techniques (Grad-CAM, LIME, SHAP) to explain model predictions.

e. Integration of Models with Different Systems
1. Lead the integration of deep learning models and camera systems with the drone and robotic systems, ensuring seamless communication between software components and hardware platforms.

f. On-Field Testing
1. Plan and conduct comprehensive on-field tests to evaluate model performance in real-world scenarios.

g. Research and Development
1. Stay abreast of the latest research in deep learning, computer vision, and robotics. Identify new techniques and methodologies to enhance project outcomes.

h. In-depth knowledge of Computer Vision Techniques
1. Apply concepts of multiview geometry for tasks such as 3D reconstruction, camera calibration, and stereo vision.
2. Implement optical flow algorithms for motion tracking and scene understanding.
3. Understand and apply epipolar constraints for improving stereo matching and 3D reconstruction accuracy.

2