Data Science / Engineering (scikit-learn, pandas, geopandas, dask, ray, torch, keras) , ML Ops, Agentic AI, Frontend (React + TypeScript), Django, Docker, AWS (EC2, Bucket, Sagemaker) / Azure (File Share / Net Files), Gitlab, Git

API / Framework for lineage and transformation processes of high-value data

• e.g., if data licenses has changed, the workflow can be re-run while excluding the
• prevent stale data. Many exploration projects are overlapping and the prospectivity maps cannot be traced where and how they were derived from.
• customizable schemas, robust parsing (including geologic ages) and forcing data type

Basic geoprocessing tools

• adaptors for converting geopandas vector data into xarray grids
• computational speed-up of libraries and routines via parallelisms of Dask, Ray and Joblib
• handling different filetypes: shp, gdb, (geo)tiffs including cloud-optimization,

Experimentals

• Sentence transformer-based data extraction (automatic conversion of "dirty" data)
  • automatic column mapping assignments
  • automatic commodity detection
  • automatic extraction of information such as commodities from text columns
• Experimental healpy-CNN deep model for prospectivity

• Develop scalable engine for publishing and ingesting data for various file types (csv, xlsx, gdb, shp)
• Maximize capability of open-source libraries for programmatic publishing to include file-specific features such as
  • Microsoft Excel Data Validation, Conditional Formatting, Cell Highlights, Calculated Formulas, Cell Comments
  • ESRI-specified features for GDB files such as Field Domains (usable in QGIS and ArcGIS)

• Docker, React JS, Vite
• Bug fixes and New Features
• Quality Assurance
• Experimental Frontend Testing Automation via Playwright

Manifold Learning on Deep Model Embeddings

• Visualize machine output (multi-dimensional embeddings) into 3D, human-understandable classifications
• Automatic highlight and labelling of classifications (via clustering model)

Feature Engineering of physically unavailable features

• regression model selection and development with scikit models and custom transformer-based deep network
• model explainability with model condensation and Accumulated Local Effects or ALE

Prospectivity model

• Automate field specialists' traditional method
• Develop and select unsupervised clustering models from neighbor-based and density-based algorithms such as OPTICS, DBSCAN, and HDBSCAN
• Automatic feature redundancy & correlation detection from Mutual Information heat maps + recursive feature elimination
• Develop a composite model from the three methods (OneClass classification, unsupervised clustering, and field specialists') to generate a prospectivity map

Others

• Refactor Paleo data calculation and aggregation code that use pygplates
• Dataiku

Courses taught:

• Chemical Process Dynamics and Control
• Mathematical Methods in Engineering / Mathematical Methods in Chemical Engineering II
• Chemical Engineering Thermodynamics Laboratory
• Introduction to Bioprocess Engineering
• Chemical Engineering Process Analysis II
• Elementary Chemical Engineering

2016-2017 Process Engineer for Thermal Powerplant and Utilities

• Cooling Water Treatment System Upgrade: identified gaps from best practices and comprehensive project proposal
• Established automated analysis for Instrument Air & Cooling Water Data
• Conduct trainings on TPP-Utilities for new hires

2017-2019 Process Engineer for Atmospheric Pipestill

• Identified and Solved Low Distillate yield in Atmospheric Pipestill due to Severe Overhead Corrosion; designed tray specifications and material type
• Case Study for economics of Salvaging of Shuaiba Refinery
• Streamlined and Established Plant-wide Workflow for Crude Oil Processability Study
• Established and organized Process Engineering Group's data folder and standardized pre-inspection preparations and checklists
• Conduct trainings on Atmospheric Pipestill process and update training materials
• Simulations in Aspen HYSYS and PRO-II for Dew Point Calculation and Overhead Washwater Rate calculations

Day-to-day tasks

• Monitoring Process conditions and equipment health
• Make recommendations for process parameter adjustments
• Field rounds
• Supervision of process-related activities
• Scheduled duty as standby firefighter

Special tasks

• Shifting schedule coverage for start-up and shutdown activities (including emergency shutdowns), including for other/units such as
  • Vacuum Pipestill
  • Gas Oil / Diesel Hydrotreaters
  • Naphtha Hydrotreaters
  • LPG Treaters
  • KeroMerox
  • Fluidized Catalytic Converter (FCC) by UOP
• inspection during vessel opening, analysis and reports
• sign off / approve vessel cleanliness and internals completion prior to closing
• field supervision and coordination for process-related activities such as
  • Catalyst loading/unloading
  • Vacuum line oxidation & declogging
  • wash water specialized nozzle tip installation
  • Passivation