Computer Vision Engineer

Spraxa Solutions; Dadri; On-site; INR 800,000 - 1,400,000; Full time

Job summary

Spraxa Solutions in India seeks a Computer Vision Engineer to train CV models and optimize them for CCTV applications. You will collaborate with CV/Deep Learning and Data Science teams to curate datasets, tackle challenging optimization tasks, and implement SOTA architectures for enhanced accuracy.

Required hands-on experience with TensorFlow/PyTorch for detection, classification and segmentation, dataset quality, training dynamics, CNNs, and regularization.

Qualifications

• Proficiency in training Detection, Classification and Segmentation models using TensorFlow or PyTorch.
• Strong understanding of dataset quality considerations for CV applications.
• Deep knowledge of model training dynamics and how to address errors from training/evaluation metrics.
• Solid grounding in CNNs and regularization techniques.

Responsibilities

• Collaborate with CV and Deep Learning teams to train CV models effectively.
• Coordinate with Data Science to curate datasets for training purposes.
• Tackle challenging problem statements to fine-tune models on large datasets.
• Implement SOTA architectures for model training and work with R&D to improve accuracy, esp. for CCTV applications.

Skills

Detection models

Classification models

Segmentation models

Model evaluation metrics

CNNs

Regularization techniques

Tools

TensorFlow

PyTorch

YOLO-series

EfficientNet/EfficientDet

Docker

Job description

Roles and Responsibilities of a Computer Vision Engineer (AI):

1. Collaborate actively with the Computer Vision and Deep Learning Team to train Computer Vision Models effectively.
2. Coordinate closely with the Data Science Team to curate appropriate datasets for training purposes.
3. Tackle challenging problem statements to fine-tune models using extensive datasets.
4. Implement State-of-the-Art (SOTA) architectures for model training purposes. Work closely with the Research and Development (R&D) Team to enhance model accuracy and precision, particularly for CCTV cameras.

Requirements:

1. Proficiency in training Detection, Classification, and Segmentation Models using TensorFlow/PyTorch.
2. Strong understanding of dataset quality considerations for Computer Vision Applications.
3. In-depth knowledge of Model Training Dynamics, including the ability to identify and address errors based on training/evaluation metrics.
4. Sound theoretical and practical grasp of Deep Learning fundamentals, such as Convolutional Neural Networks (CNNs), Regularization Techniques, etc.
5. Familiarity with cutting-edge models like YOLO-series, Efficient Net/Efficient Det, etc.
6. Experience in utilizing Docker containers for Computer Vision and Deep Learning tasks.

Good to Have:

• Experience in employing Model Optimization techniques like Pruning.
• Familiarity with FP16/Mixed Precision and INT8 Optimized Training methodologies.
• Experience: 1 to 2 years