Engineer building data, simulation & AI systems - Ph.D. (UT Austin), 10+ years of Python, energy domain.
I've spent fifteen years building the models, data systems, and now AI pipelines that turn subsurface measurements into decisions. Across every role, the real deliverable has been software: inversion code in my Ph.D., a mesh-free solver in my postdoc, a well-data platform backend at a national oil company, and now production LLM-agent pipelines.
📍 Coppell, TX (Dallas–Fort Worth)
-
🤖 Production LLM-agent pipeline for well-data QC. It is shipped as a Claude Code plugin (three chained skills) and does the following:
- multi-jurisdiction document retrieval
- grounded extraction with per-document provenance
- an append-only evidence store
- deterministic SQL conflict resolution (no LLM in the tie-break)
- geometry validation and staged human sign-off
It has been validated on ~500 wells.
-
🔗 Acquisition deal-integration pipeline. A quarterly Python pipeline that drives a reserves-software REST API end to end: entity resolution against PostgreSQL, an identifier crosswalk, and automated underwriting-vs-booked reconciliation.
-
🗄️ Sand-Control Failure Database (private). Sole data engineer for a UT Austin research consortium: schema design, ingestion, and query system, with data contributed by 10 major operators.
-
📚 Data Science Guide. A cheat-sheet site covering statistics, Python, ML, big data and deep learning, with one new sheet each week.
| Project | What it is | Stack |
|---|---|---|
production-data-analysis · prodpy on PyPI |
Production forecasting toolkit: vectorized Arps decline-curve fitting with uncertainty sampling, multi-zone production allocation, one-page production dashboards | Python, NumPy, SciPy, pandas, Matplotlib |
formation-evaluation (pphys) |
Well-log interpretation: LAS file QC and PDF reports (LasView), interactive Bokeh log viewer, one-page well and cross-section views, shaly-sand porosity and saturation models |
Python, Bokeh, Matplotlib, uv |
| wellx-webapp | Web app for field data: map-based well explorer, time-series and decline dashboards, petrophysics and flow modules behind a validated API | FastAPI, Pydantic, Vue, Leaflet, Vite |
| data-science-guide | Static cheat-sheet site with automated catalog generation, link checks, and GitHub Pages deploys | HTML/CSS/JS, Python, GitHub Actions |
- Ph.D., Petroleum Engineering - The University of Texas at Austin (advisor: Mukul Sharma). Built the forward electromagnetic simulators and the simulated-annealing inversion for a DOE-funded fracture-diagnostics tool.
- Postdoc, UT Austin. Boundary-element / integral-equation solver as a fast, mesh-free alternative to FEM/FDM for real-time flow modeling.
- SOCAR Upstream - Field Development Lead & Data Management. Wrote the entire backend of a Vue/FastAPI/PostgreSQL well-data platform used by 50+ engineers, and built dashboards that 80% of the reservoir team adopted.
- Six years teaching reservoir simulation and petrophysics (Baku Higher Oil School, METU NCC). Supervised 50+ theses.
- B.Sc. & M.Sc., Middle East Technical University. Ranked first in the department.
Domain tools: tNavigator · CMG · Eclipse · Petrel · Techlog · ComboCurve · Enverus · Spotfire
- Shiriyev et al. (2018). Experiments and simulations of a prototype triaxial electromagnetic induction logging tool for open-hole hydraulic fracture diagnostics. Geophysics, 83(3), D73–D81.
- Zhang, Shiriyev et al. (2019). Fast inversion of downhole electrical measurements for proppant mapping using very fast simulated annealing. Geophysics, 85(1).
- Shiriyev et al. (2023). Evaluating the optimal logging suite for sandstone reservoirs in the South Caspian Basin. SPE Caspian Technical Conference.
Full list: Google Scholar
Cards are rebuilt daily by a GitHub Action from public repository data.