ML Engineer - Computer Vision

Soothsayer Analytics; Hyderabad; On-site; INR 1,200,000 - 1,800,000; Full time

Job summary

Soothsayer Analytics in Hyderabad seeks an experienced Computer Vision Engineer to develop and train models for image classification, detection and segmentation. You will work with large image datasets, perform data cleaning, annotation support, and augmentation to improve accuracy.

Ideally you will have hands-on PyTorch or TensorFlow experience, a solid grounding in CNNs and model evaluation, and a strong ability to write clean Python for data processing and model development in a fast-paced

Qualifications

• Hands-on experience with at least one deep learning framework (PyTorch or TensorFlow).
• Recent experience working on Computer Vision use cases.
• Understanding of core computer vision concepts (CNNs, image preprocessing, model evaluation).
• Ability to write clean Python code for data processing and model development.
• Strong willingness to learn and work in a fast-paced environment.

Responsibilities

• Develop and train computer vision models for image classification, object detection, and segmentation.
• Work with image datasets, including data cleaning, annotation support, preprocessing, and augmentation.
• Apply deep learning frameworks to build and improve model performance.
• Implement and test anomaly detection approaches including PatchCore; work with SAM and Swin Transformers.
• Apply ML techniques to integrate visual data with structured metadata for business insights.
• Evaluate model performance using metrics like mAP, precision, recall, and F1-score; follow validation approaches.
• Assist in integrating models into existing pipelines and support deployment efforts.
• Collaborate with cross-functional teams to communicate results to technical stakeholders and clients.

Skills

Image processing pipelines

Deep learning

Model evaluation

Python

Communication with stakeholders

Education

Bachelor's degree in Computer Science/Engineering/Math

Tools

PyTorch

OpenCV

scikit-image

Detectron2

MMSegmentation

anomalib

ResNet

Faster R-CNN

Swin Transformer

MAE

Job description

Job Responsibilities

• Develop and train computer vision models for tasks such as image classification, object detection, and segmentation.
• Work with image datasets, including data cleaning, annotation support, preprocessing, and augmentation.
• Apply deep learning frameworks to build and improve model performance.
• Implement and test approaches for anomaly detection, including unsupervised defect localization using PatchCore, and work with modern architectures such as Segment Anything (SAM) and Swin Transformers.
• Apply standard machine learning techniques such as Regression, Gradient Boosting, and Time-Series methods to integrate visual data with structured metadata for business insights.
• Evaluate model performance using metrics such as mAP, precision, recall, and F1-score, and follow established validation approaches to ensure model reliability.
• Assist in integrating models into existing pipelines and support deployment efforts.
• Work closely with cross-functional teams to communicate results to technical stakeholders and clients.

Technical Skills & Stack

• Computer Vision & Deep Learning (Core Focus): Libraries: PyTorch, OpenCV, scikit-image, Detectron2, MMSegmentation, and anomalib.
• Architectures: ResNet, Faster R-CNN, Swin Transformer, Autoencoders (SAE/VAE), and Masked Autoencoders (MAE).
• Concepts: Image classification, object detection, segmentation, basic CNN architectures Machine Learning & Data Science: Techniques: Supervised/Unsupervised learning, XGBoost, LightGBM, Random Forest, PCA/SVD (Dimensionality Reduction), and Cross-Validation.
• Libraries: Scikit-learn, XGBoost, Pandas, NumPy, SciPy, and Matplotlib.
• Data & Tools: Basic SQL knowledge Familiarity with Git or version control Experience working with structured and unstructured datasets

Education & Experience Experience:

• 4-5 years of overall experience, with recent hands-on work in Computer Vision projects for at least 1 year
• Education: Bachelor's degree in Computer Science, Engineering, Mathematics, or related field
• Must-Have Requirements
• Hands-on experience with at least one deep learning framework (PyTorch or TensorFlow)
• Recent experience working on Computer Vision use cases
• Understanding of core computer vision concepts (CNNs, image preprocessing, model evaluation)
• Ability to write clean Python code for data processing and model development
• Strong willingness to learn and work in a fast-paced environment .