« Back to course list

Advanced Neural Networks, Deep Learning and MLOps for Upstream Professionals (RES77)

  • 1-5 December 20245 daysDoha, QatarCourse Fee: 5350 USD
    Computer Fee: 350 USD


This highly practical course is designed to equip a participant with a comprehensive understanding and solid practical knowledge of main Deep Learning algorithms including Reinforcement Learning, Generative Adversarial Networks (GAN) and Large Language Models (LLM). Through a well-balanced combination of theory deep dives and hands-on exercises using real-world Oil & Gas datasets, a participant will acquire the necessary skills to confidently leverage new knowledge and the full potential of Deep Learning, effectively addressing day-to-day Oil & Gas problems.

  • Minimum Specifications: Participants are required to have access to modern laptops or PCs with a minimum of 8 GB of RAM and at least 30 GB of free disk space.
  • Recommended Specifications: For optimal performance, it is recommended that participants use a computer with 16 GB or more of RAM and maintain 30 GB of free disk space.

Course Level: Advanced
Instructor: Vitali Molchan

Designed for you, if you are...

  • A reservoir engineer, geologist, petrophysicist, production or drilling engineer with prior programming experience and basic knowledge of data science and machine learning who wants to obtain a solid understanding of neural networks, deep learning and machine learning operations (MLOps)

  • Participants should have strong domain knowledge with a minimum of 5 years of experience
  • Participants should have prior programming experience
  • Prior experience in data science and machine learning is an advantage. If a participant does not have basic knowledge of data science and machine learning, it is recommended to attend the following course:
    Applied Machine Learning and Data Science for Upstream Professionals.

How we build your confidence

  • The main Deep Learning algorithms will be thoroughly discussed and accompanied by numerous reusable code examples based on real-world Oil & Gas data sets
  • MLOps concepts will be discussed to guide the development of comprehensive end-to-end ML solutions including tasks from project scoping to model serving and building a graphical user interface

The benefits from attending

By the end of the course, you will get a robust comprehension of:

  • Opportunities identification for applying Deep Learning methods in your professional domain
  • Sound decision-making in selecting the most suitable machine learning methods for solving specific problems
  • Core Deep Learning algorithms and their implementation using TensorFlow and Keras libraries
  • Practical application of main machine learning methods in real-world Oil & Gas scenarios


Day 1:
  • Introduction to Deep Learning and TensorFlow ecosystem
  • Machine learning 101
  • Introduction to MLOps
  • Deep Feed Forward neural networks

  • Machine learning project setup
  • Production forecast for unconventional reservoirs
  • Reservoir fluids saturation pressure forecast
  • Lithofacies identification

You will learn how to:
  • Set up a machine learning project to ensure the reproducibility and model accuracy
  • Use version control, docker, and code templates to streamline the model development and deployment into production
  • Construct Deep Feed Forward neural networks using the TensorFlow and Keras framework
  • Use various approaches to improve the accuracy of the Deep Learning models
  • Explain the prediction of the Deep Learning models

Day 2:
  • Time-series modelling
  • Anomaly detection
  • Bayesian neural networks and Probabilistic Forecasting
  • MLOps: data pipelines, hyperparameters tuning and tracking modelling experiments

  • Predictive maintenance of the Electrical Submergible Pumps (ESP)
  • Production forecasting for the unconventional reservoirs
  • Anomalies detection in production well behaviours

You will learn how to:
  • Improve utilization of the Electrical Submersible Pump (ESP) using predictive maintenance
  • Detect non-stationary production regimes in production and injection wells
  • Perform a probabilistic production forecasting for unconventional oil and gas reservoirs
  • Build robust data pipelines to ensure the Deep Learning model quality
  • Define the best architecture of the Deep neural network, perform hyperparameters tuning and track the modelling experiments
  • Serve the Deep Learning models in production

Day 3:
  • Deep convolutional neural networks
  • Object detection models
  • Image segmentation models
  • MLOps: Deep Learning model serving

  • Sucker-rod pump performance monitoring
  • Microfacies classification based on thin section images
  • Object detection based on satellite images

You will learn how to:
  • Monitor the sucker-rod pump performance using automatic interpretation of the dynamogram data
  • Detect petrographic microfacies based on thin-section images
  • Detect wells, pipelines, tanks, oil spills and other objects of interest in the satellite image data

Day 4:
  • Text classification
  • Named Entity Recognition (NER)
  • Text summarization
  • Large Language Models (LLM)
  • MLOps: building the interface for a deep learning model and dashboarding

  • Developing Oil & Gas word embeddings
  • Classification of journal articles
  • Creating a Named Entity Recognition (NER) model to detect Oil & Gas-related vocabulary
  • Generating abstracts to summarize the journal articles

You will learn how to:
  • Classify the text documents (reports, articles, etc.) based on certain topics
  • Identify a certain type of information within text documents
  • Create a summary of large text documents
  • Extract the knowledge for a large number of unstructured text documents
  • Use pre-trained Large Language Models for NLP modelling
  • Build a graphical user interface (GUI) to simplify deep learning model usage

Day 5:
  • Reinforcement learning
  • Generative AI: Generative Adversarial Networks (GANs)

  • Optimal well placement
  • Generating synthetic well data

You will learn how to:
  • Perform the optimal well placement using reinforcement learning
  • Generate synthetic well data (well logs, core images, sensor data, etc.) to improve downstream model accuracy



    © All rights reserved
    HOT Engineering GmbH   Tel: +43 3842 43 0 53-0   Fax +43 3842 43 0 53-1   hot@hoteng.com