Industrial AI for Energy Operations | Data Scientist | Completions & Well Intervention Engineer | Author
I build AI and data systems for energy operations — combining 17+ years in completions and well intervention across multiple countries and operators with an MSc in Data Science applied to real engineering problems, not generic modelling exercises. My focus: retrieval-augmented generation, operational report intelligence, and evidence-backed decision support for drilling and completions workflows.
Building Industrial RAG Systems from Daily Drilling Reports
A free, hands-on book teaching engineers to build retrieval-augmented generation systems directly on Daily Drilling Reports — written for fellow drilling and completions engineers to apply to their own workflows, not as a commercial product. It covers:
- RAG, from keyword search through hybrid retrieval
- Engineering knowledge extraction from unstructured operational text
- Systems where every generated answer cites its source
- Lessons-learned extraction and human-in-the-loop review workflows
- Practical implementation grounded in a real, public well archive
🔭 Currently building: the DDR Intelligence Platform (drilling report semantic search, sequence mining, NPT precursor detection) and Completion Campaign Intelligence (NLP-driven completion/stimulation campaign analysis) — both in Featured Projects below, alongside my public and open-source work.
- Languages: Python · R · SQL
- AI & Retrieval: RAG · FAISS · Hybrid Search (BM25 + Dense) · LLM Evaluation
- Data Science: pandas · NumPy · scikit-learn · Statistical & Time Series Analysis
- Platforms: Dataiku · Streamlit · Quarto · Jupyter
- Tooling: Git · GitHub · PyCharm
| Project | What it demonstrates | Stack | Access |
|---|---|---|---|
| EvidenceRAG-Evaluation | MSc thesis: hybrid retrieval evaluation harness (BM25 + dense + Reciprocal Rank Fusion) for grounded, citation-backed QA, with a hallucination-reduction focus | Python · FAISS · Sentence-Transformers · PyTorch | Public |
| Frac_Campaign_Planning | Monte Carlo simulator for multi-pad hydraulic fracturing campaign scheduling, risk, and scenario optimisation | R · Shiny | Public |
| GP_Screens_Analysis | Computer vision pipeline detecting, classifying, and quantifying failure modes on failed gravel pack screens | Python · OpenCV · YOLO (Ultralytics) | Public |
| AI Home Energy Intelligence Platform | Local-first, deterministic home energy analyst — weather-adjusted regression, 8-model forecasting, anomaly detection, and an evidence-cited Q&A consultant, no cloud or LLM required | Python · Streamlit | Public |
| ccs-workover-forecast | Reliability-driven Monte Carlo simulator for CCS well workover and intervention demand forecasting | Python · SciPy · lifelines · Streamlit | Public |
| DDR Intelligence Platform | AI-powered drilling report analytics: sequence mining, NPT precursor detection, and semantic search | — | Private — available on request |
| Completion Campaign Intelligence | NLP-driven intelligence platform for completion and stimulation campaigns | Python · NLP · Local LLM (Ollama) | Private — available on request |
Oil & Gas: Well Intervention · Completions Engineering · Sand Control · Hydraulic Fracturing · Artificial Lift · Workovers · Well Integrity · Decommissioning · CCS Wells
Data Science & AI: Machine Learning · Predictive Modelling · Time Series Analysis · Retrieval-Augmented Generation · Hybrid Search · Knowledge Extraction · Information Retrieval · Operational Analytics
- Industrial / Energy AI collaborations — RAG, knowledge extraction, and operational-intelligence projects built on real reports and field data
- Data science / applied AI roles in energy, operations, or engineering analytics
- Technical writing, applied-AI education, and advisory work
🌍 Scotland, United Kingdom
Industrial AI for energy operations — turning field experience and engineering reports into traceable, evidence-backed intelligence.