Senior Data Scientist with PyTorch

EPAM Systems; United States; Remote; USD 120,000 - 180,000; Full time

Job summary

EPAM Systems in the United States is seeking a Senior Data Scientist with PyTorch expertise to join a cross-disciplinary team of geoscientists and ML engineers working on seismic data. You will translate technical and business needs into reliable deep learning solutions, with emphasis on denoising, segmentation, and structural interpretation using 2D/3D seismic imagery.

The role involves improving existing seismic ML solutions, retraining models, and exploring new use cases across exploration

Qualifications

• 2–4 years of experience in ML/DS roles with deep learning on image or volumetric data
• Knowledge of computer vision including segmentation and feature extraction
• Proficiency in PyTorch for building and evaluating models
• Experience with 2D/3D seismic data and architectures like U-Net and CNNs
• Strong Python skills
• English proficiency at B2 level or higher
• Familiarity with MLOps tools and cloud services

Responsibilities

• Apply and develop deep learning models for denoising and structural interpretation of seismic data
• Improve the accuracy, robustness and performance of existing seismic ML solutions
• Develop models for structural interpretation tasks such as fault detection, horizon tracking and seismic facies classification
• Collaborate with Geoscientists and Geophysicists to validate model outputs against expert interpretation and well data
• Identify, evaluate and propose new business use cases for computer vision and deep learning across workflows
• Stay up to date with research in computer vision and seismic ML
• Follow good ML engineering practices for reproducible experiments and model deployment

Skills

PyTorch

Python

Deep learning

Computer Vision

Seismic data

Tools

MLflow

Docker

Cloud services (Azure/AWS/GCP)

Job description

We are seeking a Senior Data Scientist with PyTorch expertise to join a team of Geoscientists, Geophysicists, Data Scientists and Machine Learning Engineers, working to understand technical and business needs and translate them into reliable, high-impact deep learning solutions for seismic data. This role will also focus on improving existing seismic ML solutions and identifying new business use cases.

Responsibilities

• Apply and develop deep learning models (U-Nets, CNNs and related architectures) for denoising and structural interpretation of 2D and 3D seismic data
• Improve the accuracy, robustness and performance of existing seismic ML solutions, including model retraining, architecture refinement and hyperparameter tuning
• Develop models for structural interpretation tasks such as fault detection, horizon tracking and seismic facies classification
• Work closely with Geoscientists and Geophysicists to validate model outputs against expert interpretation and well data, and incorporate their feedback into model improvements
• Identify, evaluate and propose new business use cases where computer vision and deep learning can add value across exploration and production workflows
• Stay up to date with research in computer vision and seismic ML, and assess new methods for practical application
• Follow good engineering practices for reproducible experiments, model versioning and deployment of models into production workflows

Requirements

• Preferably 2-4 years of experience in similar roles (ML Engineer, Data Scientist or Geophysicist with a strong ML focus), including hands-on experience applying deep learning to real-world image or volumetric data
• Understanding of Computer Vision, including image segmentation, denoising, feature extraction and convolutional architectures
• Proficiency in PyTorch for building, training and evaluating deep learning models
• Expertise in U-Net, CNN and related architectures for 2D and 3D seismic data
• Background in the Upstream Oil & Gas domain, working with Geoscience and Seismic data and teams
• Understanding of seismic data concepts such as acquisition and processing basics, seismic attributes, noise types and structural features (faults, horizons)
• Strong Python skills
• English proficiency at B2 level or higher
• Nice to have Experience with transformer-based or self-supervised models for vision or seismic data
• Familiarity with MLOps tools (e.g. MLflow), Docker and cloud AI services (Azure, AWS or GCP)