Enterprise Productivity for Upstream Operators

Turn fragmented engineering work into faster field decisions.

Groundwork maps the work between drilling, completions, reservoir, and production teams — then builds the private systems that make high-value engineering workflows repeatable.

No credentials, exports, or direct system access required for the first call. For VP Engineering, CTO, Reservoir, Production, and Asset leaders at independent E&P operators.

Built by Dr. Mehrdad G. Shirangi — Stanford PhD · former AI Transformation & Technology Leader, Baker Hughes · former Head of AI Innovation, Manufacturing Operations, Cisco Supply Chain Groundwork on LinkedIn →

Built on petropt — 50+ functions, MIT · petro-mcp — 70 petroleum engineering tools · tools.petropt.com — petroleum engineering workflow suites

The bottleneck is not talent. It is fragmented work.

Your engineers know what to do. Too much of their time is lost finding context, reconciling assumptions, rebuilding analyses, and translating work across teams.

Analyst hours disappear into repeat work.

Manual data pulls, spreadsheet rebuilds, status updates, sanity-check loops. Highly-paid engineers doing low-leverage work because the system makes them.

Decisions get trapped in silos.

Production, drilling, completions, and reservoir each hold partial context. By the time the right person sees the right number, the decision window is closed.

Technical workflows don’t compound.

Every project restarts from scratch. The decline-curve QA loop, the type-curve refresh, the production morning report — rebuilt by a new engineer every six months instead of becoming reusable infrastructure.

What we work on.

If your team recognizes any of these, the call will be useful.

The handoff between completions and reservoir.

The week-three production diagnostic.

The decline-curve QA loop.

The Monday-morning forecast call.

The reserves audit binder rebuild.

The morning rate-volume reconciliation.

Each one is a workflow that quietly costs your team hours every week. We turn them into systems that don’t.

Private Briefings

Three operator bottlenecks we don’t show publicly.

Each is a workflow pattern Groundwork has built for operators behind an NDA. On the briefing, we walk through one relevant to your team. No credentials, exports, or system access required.

The week-three production diagnostic

The morning rate-volume reconciliation

The decline-curve QA loop

Request a Confidential 30-Min Briefing →

What’s public is the infrastructure. The work we do for operators is bespoke and stays between us.

Open-source primitives anyone can use. Engagements built for one operator at a time, behind an NDA.

Public · Open Source

The infrastructure layer.

  • petropt — the petroleum engineering Python library
  • petro-mcp — the MCP server that connects it to AI assistants
  • tools.petropt.com — 10 engineering workflow suites by discipline (paid; sample demos public)
  • Standard methods, transparent code, MIT-licensed

petropt and petro-mcp are MIT-licensed open source and stay free. tools.petropt.com offers public sample demos; full suite access is paid.

Enterprise · Bespoke Engagements

What we build for operators.

  • Workflow systems aimed at analyst non-productive time
  • Cross-functional decision capture across production, drilling, completions, reservoir
  • Data and context integration without rip-and-replace
  • AI agents inside engineering workflows your team already runs
  • Engineering standards and adoption patterns that don’t depend on a single champion

Founder-led. Mutual NDA before any data is reviewed. Scoped from a 30-minute conversation.

Built on a decade of domain-specific engineering work. The open-source layer is the tip; the rest is bespoke per operator. Patterns don’t generalize from a screenshot.

How we engage.

One arc, five steps: define the decision → integrate the context → automate the workflow → embed the AI → sustain the standard. Most engagements move through it. Sponsored open source sits alongside.

Step 1 · Define

Decision System Design

For recurring technical decisions that need shared context, traceability, and consistent review. Replace tribal-knowledge meetings with a process the team can run.

Step 2 · Integrate

Data-Context Integration

For workflows where production, completions, drilling, reservoir, and economics data have to meet in one place. No rip-and-replace.

Step 3 · Automate

Workflow Automation

Take repeat engineering work that shouldn’t require manual rebuilds every week and turn it into a system the team can rely on.

Step 4 · Embed AI

Engineering AI Enablement

Practical AI agents inside real petroleum workflows — not generic chatbots, not disconnected pilots. Built on the same open-source primitives we maintain publicly.

Step 5 · Sustain

Team Adoption & Standards

Make the workflow usable by your engineers, not just the champion who sponsored it. Onboarding paths, review checklists, light-weight governance — so the work doesn’t depend on a single person.

Alongside

Sponsored Open-Source Contributions

Pay to add a capability to petropt or petro-mcp that benefits your team and the wider community. Your name in the changelog. The code stays MIT-licensed.

From first workflow to production system.

1

30-Minute Fit Call

We identify the workflow, the decision process, and your team context. No private system access required. Either answer — fit or no fit — is useful.

2

Focused Scoping

NDA in place. We map the current workflow, quantify the friction, and define what a useful first deployment would need to prove. In writing — what gets built, what success looks like, who owns what.

3

Pilot or SOW

Build around a narrow, high-value workflow with clear technical owners, success criteria, and handoff expectations. Code, models, and documentation hand back to your team. No vendor lock-in.

Who’s Behind It

Dr. Mehrdad G. Shirangi · Founder

Stanford PhD in energy systems optimization. Former AI Transformation & Technology Leader at Baker Hughes. Former Head of AI Innovation, Manufacturing Operations at Cisco Supply Chain — deploying AI at scale across one of the world’s most complex global hardware supply chains. Founded Groundwork in 2018 to bring AI, generative AI, and now AI agents to oil & gas. The model: founder-led from problem to deployment — Mehrdad sets the technical direction and engages directly with operators, with a hand-picked network of senior engineers and specialists brought in per scope when an engagement needs them. No slide-deck consulting, no anonymous hand-offs. Maintainer behind petropt and petro-mcp — the open-source layer Groundwork builds engagements on top of.

How We Handle Your Data

Mutual NDA before any data review. Work runs in your cloud or on dedicated infrastructure for the engagement. For consulting work, data handling is governed by the SOW or a separate Data Processing Agreement, with terms agreed mutually before any data is shared.

Built for Operators, Not Supermajors

Engagements are sized for independent E&P operators and PE-backed teams that want engineering work to move faster without turning every workflow into a custom internal software project. Asset scale qualified privately on the call.

Common Questions

Why no client logos?

Most operator work is NDA-bound. Our public proof is code, tools, and the founder’s technical track record — not a logo wall. GitHub, PyPI, and tools.petropt.com are all you need to verify the engineering depth before we ever talk.

Do you need access to our private systems for the first call?

No. The first conversation is about workflow fit: where engineering time is being lost, which decisions are recurring, and what systems your team already uses. Integration scope is defined later.

Is this software or consulting?

Neither. We build software systems for one operator at a time, on a fixed scope. The output is working code and a workflow your team owns — not a report, not seat licenses, not an ongoing retainer dependency.

Can you show demos publicly?

Most demos are private after a fit conversation, because the highest-value workflows involve company-specific data, operating context, and internal decision processes. The public open-source layer (petropt, petro-mcp) gives you a fair read on the technical depth before we ever talk.

Do you replace our existing engineering tools?

Usually no. Groundwork is designed around the workflows your team already runs — the spreadsheets, databases, production systems, technical reviews, and decision meetings where work currently fragments. We integrate, automate, and standardize. We don’t rip-and-replace.

Do you sign NDAs and DPAs?

Yes. Mutual NDA before any technical conversation that touches your data. For data processing, we work to your DPA template or a standard one we provide. We can work through your security review questionnaire as part of scoping.

Who owns the IP we pay for?

IP and licensing terms are written into each SOW. Typical scoping starts from: you own deliverables built on your data; improvements to the open-source petropt library remain MIT-licensed and benefit the wider community. The exact split is agreed in writing per engagement — no ambiguity, no surprise carve-outs.

How is this different from generic AI consulting?

Groundwork starts from upstream engineering workflows, not generic automation opportunities. The open-source layer proves the technical foundation; the enterprise work adapts that foundation to your private operating context. We don’t do AI strategy decks, AI readiness scores, or AI maturity frameworks.

Who is this for?

VP Engineering, CTO, Reservoir, Production, Asset, and Technical leaders at independent E&P operators and PE-backed teams who want engineering work to move faster without turning every workflow into a custom internal software project.

Find the highest-value workflow to fix first.

Bring one recurring workflow that costs your team time, slows a decision, or depends too much on one person’s spreadsheet. We’ll help determine whether it’s a fit for an enterprise engagement.

No credentials, exports, or direct system access required for the first call.