Over 500+ hours of learning about the latest in the field of data science, with an emphasis on learning via doing labs (over 20 labs) and projects (a total of five course projects with 1 capstone project) in a collaborative learning and work environment. Summary of the projects:- Airbnb market in Austin, TX: Investigated the Airbnb market in Austin and developed an application to help investors estimate yearly income for their home and identify investment opportunities.- Reddit post classification: Used NLP, sentiment analysis, and word transformers (CountVectorizer and TfidVectorizer) to classify subreddit posts.- Global corruption: Performed extensive web search, data extraction, data cleaning, and modeling to identify factors that influence corruption in different regions of the world.- Hotdog-not-hotdog: Used convolution neural network for image processing and classification of different food images!

- Created a Python package to streamline data processing for team members, enhancing productivity and code reusability.- Led a group of 10 developers to create and deploy a cloud app for well data analysis, reservoir modeling, and data visualization. The app was successfully deployed on AWS and Azure and was estimated to reduce the customer’s operating cost by 30% through an all-in-one solution and subscription-based pricing. - Improved delivery cycle time and resilience using DevOps and Agile practices (e.g. CI/CD and automated testing).- Collaborated with internal and external stakeholders to prioritize features and allocate resources.

- Leveraged different ML algorithms (including ANN and tree-based regression) to predict fluid production rates from underground reservoirs where factors like fluid/rock properties and operating conditions were used as features. The presented proxy workflow was deployed on two different fields and reduced the computational cost by 80-90% during field development planning. - Performed exploratory data analysis (EDA) and data wrangling on ~3000 oil/gas wells with high-frequency data (~1 million records reach). Implemented different unsupervised algorithms to cluster the data and reduce dimensionality (including K-means, DBSCAN, and PCA). Results used to guide a giant-field simulation conversion project where the end product provided a 30-70% reduction in modeling time and operating cost.- Conducted long-term CO2 storage and risk management using data analytics, numerical modeling, and invasion percolation theory. Linear regression and XGBoost were used for predictive modeling of long-term CO2 storage, identifying risks, and determining variable importance. The presented workflow reduces the decision-making time for carbon capture and storage projects.

- Multiple porosity approach to simulate fluid flow in shale by decomposing shale matrix into different porosity systems of distinctive hydraulic and transport properties.- Improved phase behavior and enhanced recovery from low permeability shale reservoirs by developing a compositional 3D simulator for multiphase fluid and heat transport in geologic media - Analyzed hydraulically fractured well test data of shale gas reservoirs in Haynesville shale (methodically classify multiple wells based on Rate Normalized Pressure (RNP) data)

- Utilized different clustering and unsupervised learning algorithms for petrophysical data analysis, rock typing, and flow simulation in a closed-loop end-to-end workflow. - The presented methodology resulted in a proposed $5-10MM cost-saving plan (by timing the next seismic surveys) and a U.S. Patent is filed based on the results of this project.

- Carried out data cleaning and uncertainty assessment on well production data from different CO2 storage sites. - Pioneered a workflow for using data analytics and numerical modeling to monitor underground carbon storage processes. Presented insights towards ways of detecting CO2 leakage during early stages.- Analyzed CO2 de-facto storage during CO2 EOR/storage processes through a compositional approach.- Performed uncertainty assessment studies for various underground CO2 injection projects.

- Simulated CO2 injection process in Cranfield Field, southwest Mississippi.- History matching of a coupled CO2 sequestration and enhanced oil recovery process.- Application of 4D seismic method to monitor gas saturation in a field study CO2 sequestration project. - Developed a new way to monitor above-zone monitoring intervals using transient well test analysis.

Utilization of Ionizing electron particles as an efficient way to reduce the viscosity of heavy petroleum fluids

Heavy oil recovery techniques

Drilling mud analysis

Routine and Special Core Analysis (SCAL)