Editorial disclosure
This article reflects the independent analysis and professional opinion of the author, informed by published research, public statements, and industry data. No company reviewed or influenced this content prior to publication. All quotes are sourced and attributed. Where predictions are speculative, they are labeled as such. Where we reference Groundwork Analytics projects, we say so explicitly.
I need to tell you something that your career services office probably will not. Not because they are hiding it, but because they do not fully understand it yet.
The tasks that were supposed to be your first job — the ones that every petroleum engineer before you spent two years grinding through — are being automated right now. Not in five years. Not when some sci-fi future arrives. Right now, in 2026, with tools you can download today.
Claude Code writes Python scripts for production data analysis. Cursor generates well performance dashboards from a text prompt. OpenClaw and NemoClaw run autonomous production surveillance and generate morning reports without a human touching them. SLB's Tela platform deploys AI agents that interpret well logs, predict drilling issues, and optimize equipment — tasks that used to be how you learned the business.
An SPE technical paper published this year described these AI agents as systems that "reason, adapt, and collaborate, like a junior engineer who understands the tools, knows when something looks wrong, and can explain the process."
Read that again. Like a junior engineer. That is you they are describing. That is the role they are automating.
This is not an article about how AI is going to end petroleum engineering. It is not. The physics is real, the wells are real, and the industry needs tens of thousands of workers to replace retirees over the next decade. But the path from graduation to competent engineer — that path is being fundamentally reshaped. And if you are a PE student or new grad, you need to understand how, and what to do about it.
The New Reality: Your First Job Looks Different Now
Let me be specific about what has changed, because vague warnings about "AI disruption" are useless. Here are the tasks that constituted the first two years of a petroleum engineer's career, and what is happening to each of them.
Production data analysis. You used to spend weeks pulling data from SCADA systems, cleaning it in Excel, building decline curves, and formatting it into reports your manager could present. Now? An AI agent connected to a production database through MCP (Model Context Protocol) can pull, clean, analyze, and visualize that data in minutes. Not hours. Minutes. Tools like petro-mcp — one of the first MCP servers built specifically for petroleum engineering data — demonstrate what this looks like in practice. The junior engineer who used to spend a week on this does not have a week of work anymore.
Decline curve fitting. This was bread-and-butter entry-level work. Pull the production history, fit Arps curves, maybe run some probabilistic forecasts. Physics-informed machine learning now does this faster and, in many cases, more accurately than manual fitting. It handles thousands of wells simultaneously. The value is no longer in doing the fit — it is in knowing what the fit means and what to do about it.
Report generation. Morning reports, weekly production summaries, drilling performance updates. Companies are deploying AI agents that generate these automatically from real-time data feeds. Baker Hughes' Leucipa platform, SLB's Lumi agents, and smaller tools built on Claude and GPT handle report generation as a background task. The engineer who compiled these reports is being replaced by a cron job.
Drilling parameter monitoring. Watching real-time drilling data for anomalies, adjusting parameters, flagging issues for the drilling engineer — this was a classic junior role on the rig or in the operations center. Real-time AI optimization systems now monitor hundreds of parameters simultaneously and flag issues before a human would notice them. ExxonMobil reduced well planning time from nine months to seven months with AI tools. The monitoring that used to train you is being done by software.
Completion design comparisons. Evaluating offset wells, comparing frac designs, analyzing stage-by-stage data — ML handles thousands of wells at once, across basins, with statistical rigor that no spreadsheet can match. The junior engineer who spent months building comparison databases is competing with tools that build better ones overnight.
I am not listing these to scare you. I am listing them because you need to know exactly what is being automated so you can position yourself above it, not in front of it.
What Used to Be Your Path (And Why It Is Breaking)
The traditional petroleum engineering career path looked like this:
Graduate with a BS or MS in petroleum engineering. Land an entry-level role — production engineer, reservoir analyst, field engineer, operations support. Spend two years doing grunt work — data pulls, Excel models, basic analysis, running simulations someone else designed, sitting in on meetings where you mostly listened. Learn the business through that grunt work — understanding why the numbers matter, what the patterns mean, how decisions get made. Get promoted to a role where you actually make decisions, because you earned the context through those two years of grinding.
That grunt work was not just busy work. It was the training ground. You learned what production data smells like when a well has a problem. You learned which completion designs work in the Wolfcamp A versus the Wolfcamp B because you manually compared hundreds of them. You learned how to communicate with operations because you sat next to them pulling their data every morning.
AI is removing the training ground while the industry still expects you to have the knowledge that training ground provided.
This is the core problem, and nobody is talking about it clearly enough. Companies are not eliminating the need for experienced petroleum engineers. They are eliminating the path that creates them.
Here is what this looks like in practice. A mid-size Permian operator used to hire four to six new petroleum engineers every year. Two went to the field, two went to the office, and they rotated through production, reservoir, and completions over three years. Now that same operator has AI agents handling production surveillance, automated decline curve analysis, and ML-assisted completion design. They might hire two new engineers instead of six — and those two need to hit the ground running because the ramp-up work that used to take two years is gone.
The math is brutal. Fewer entry points. Higher expectations at each entry point. And a generation of engineering students who trained for jobs that are transforming underneath them.
The Numbers Behind the Narrative
I want to be honest about what the data does and does not show, because this conversation has too much hype in both directions.
What the data clearly shows:
Entry-level job postings across all industries dropped 38% between 2022 and 2023. In software development and data analysis — the skills most adjacent to what PE graduates do — postings plummeted 67%. Big tech entry-level hiring dropped more than 50% over three years. A Harvard study found that after late 2022, companies actively adopting AI hired five fewer junior workers per quarter than before.
The underemployment rate for recent college graduates hit 42.5% in Q4 2025 — the highest since 2020. That means nearly half of new graduates are working in roles that do not require a degree. Oxford Economics estimates that 85% of the rise in U.S. unemployment since mid-2023 stems from new graduates struggling to land entry-level positions.
The Gusto New Grad Hiring Report found that the average hiring rate for the Class of 2025 was 4.8% — 44% lower than the Class of 2022.
And 60% of positions labeled "entry-level" in software and IT now require three or more years of experience. The label has become meaningless.
What the big voices are saying:
Dario Amodei, CEO of Anthropic, predicted that AI could automate 50% of all entry-level white-collar jobs within one to five years. Bill McDermott, CEO of ServiceNow, said unemployment for new college graduates "could easily go into the mid-30s in the next couple of years." Geoffrey Hinton, the "godfather of AI," warned of "massive unemployment caused by AI" and said he does not expect new jobs to come close to replacing those eliminated.
What the counter-arguments say:
The Yale Budget Lab found "no clear evidence that the U.S. workforce has undergone widespread displacement" through late 2025. The occupational mix has remained "strikingly stable." Many CEOs at Davos disagreed with Amodei's timeline. And Klarna's cautionary tale — they replaced 700 customer service agents with AI, then had to rehire humans when quality collapsed — shows that AI replacement has real limits.
My honest read:
The aggregate data has not caught up yet. Labor market statistics are lagging indicators. But the hiring freeze at the entry level is real and measurable right now. Companies are not mass-firing junior engineers — they are simply not hiring as many. The pipeline is being quietly narrowed, not dramatically cut. And for a PE student graduating in May 2026, the distinction between "not hired" and "laid off" is academic. The result is the same: you are competing for fewer spots with higher requirements.
The Paradox: More Tools, Fewer Entry Points
Here is the irony that should frustrate every engineering student.
The oil and gas industry has more AI tools than at any point in its history. The AI in oil and gas market hit $4.55 billion in 2026 and is growing at nearly 13% per year. SLB has Tela. Baker Hughes has Leucipa. Halliburton has AI-assisted drilling optimization. Every major service company and most large operators are deploying AI agents, copilots, and automation platforms.
These tools need people who understand them. People who can configure them, validate their outputs, catch their mistakes, and apply them to real engineering problems. That should be a massive opportunity for new graduates.
But companies want "2-3 years experience with AI/ML tools in upstream operations" on the job posting. They will not hire you to get that experience. They want you to already have it.
This is the chicken-and-egg problem of 2026. The industry desperately needs AI-literate petroleum engineers. Entry-level roles are where you would normally build that literacy. But the entry-level roles are being automated by the very tools you need experience with.
Anthropic's own CPO, Mike Krieger — co-founder of Instagram — admits they rarely hire fresh graduates. Anthropic has no internship program. The company building the AI that automates entry-level work does not hire entry-level workers. If that does not crystallize the problem, nothing will.
Korn Ferry's research puts it bluntly: 37% of companies plan to replace entry-level roles with AI. But those entry-level employees are foundational to the future leadership pipeline. These roles provide learning, institutional knowledge, and skill development that cannot be replicated by AI training programs. Companies saving money on junior hires today are creating a leadership vacuum ten years from now.
The consulting world is the canary in this coal mine. McKinsey's Lilli and BCG's Deckster now perform roughly 80% of the work junior analysts used to do — research, data analysis, slide generation. KPMG cut entry-level hiring by 29%. Deloitte cut by 18%. Starting salaries at top firms have been frozen for three consecutive years. The "pyramid model" that built every major consulting firm — thousands of junior analysts doing the base work so partners can bill the strategic work — is cracking.
The same dynamic is coming to oil and gas. Slower, because physical operations cannot be fully digitized, but it is coming. And unlike consulting, PE has a much smaller talent pool to begin with. When the funnel narrows, it narrows fast.
What Actually Gets You Hired Now
I am not going to give you motivational poster advice. "Follow your passion" and "network more" are not strategies when the fundamental structure of entry-level hiring is shifting. Here is what I believe actually works, based on what I see hiring managers value, what I see in the market, and what I know about how AI is changing engineering work.
Build things. Do not just study things.
A GitHub profile with real projects is worth more than a 3.8 GPA in the current market. Not because grades do not matter — they do — but because the signal has changed. When AI can pass most engineering exams, the exam grade tells a hiring manager less than it used to. What tells them more is evidence that you can take a real problem, break it down, and build something that works.
Specific things you can do:
- •Contribute to open-source petroleum engineering projects. petro-mcp is an MCP server for petroleum engineering data — it needs contributors. lasio is the standard Python library for reading LAS files. pyResToolbox provides reservoir engineering utilities. These projects are used by real engineers on real problems. Contributing to them gives you something concrete to point to.
- •Analyze public data. The Texas Railroad Commission publishes production data. FracFocus publishes completion data. The EIA publishes market data. Take a basin you care about, build a decline curve analysis pipeline, generate type curves, and publish it. A hiring manager in Midland will click on that.
- •Build a tool someone would actually use. A well economics calculator. A production dashboard. An ESP sizing tool. Something that shows you understand both the engineering and the code. Not a class project — something you built because you wanted to solve a problem.
Learn to direct AI, not compete with it.
This is the single most important shift in how engineering careers will work.
The engineer who can tell Claude Code what to build — who can write the right prompt, validate the output, catch the physics errors, and iterate toward a solution — is worth significantly more than an engineer who cannot use AI tools at all. And critically, that engineer is also not threatened by the AI, because the AI does not know what question to ask. You do.
This is not "prompt engineering" as a buzzword. It is the practical skill of combining domain knowledge with AI tools to solve problems faster than either could alone. It means:
- •Knowing enough Python to read and modify what Claude Code generates, not just accept it blindly.
- •Understanding the petroleum engineering well enough to catch when an AI tool produces a physically impossible result — a negative skin factor where the completion suggests damage, a decline curve that implies more reserves than the volumetrics support, a drilling recommendation that ignores the formation pressure.
- •Being able to articulate what you need in technical terms precise enough that an AI agent produces useful output on the first or second try, not the tenth.
The combination of domain knowledge plus AI fluency is rare. It is getting rarer as PE enrollment drops. It is the new skill stack, and it is learnable.
Get domain experience any way you can.
If companies will not give you a traditional entry-level role, find other paths to the same knowledge. This is harder than it used to be, and I will not pretend otherwise. But the paths exist.
Open-source contributions. I mentioned this above, but it bears repeating. Contributing to tools that real engineers use forces you to understand real engineering problems. You learn by building. This is the closest thing to on-the-job training that exists outside of a job.
Research collaborations. Work with faculty on applied problems — not theoretical dissertations, but projects that use real data and produce real tools. Several programs now offer relevant tracks: Texas A&M's data analytics certificate, Colorado School of Mines' PE Data Analytics program, OU's dual-degree options. If your university has industry partnerships, get involved with the applied work, not just the coursework.
Service company field roles. Someone still has to be on the rig. Wireline, MWD/LWD, frac crews — these roles still need humans, and they give you the domain knowledge that no amount of coding can replace. The field engineer who also understands data and AI tools is an extremely valuable person. The path is harder physically, but the learning is irreplaceable.
Consulting partnerships. Small firms working on real client problems sometimes need extra hands. This is not glamorous. It might not pay well initially. But it gives you real-world exposure that a classroom cannot.
Target small operators, not supermajors.
A 200-well company in Midland, a PE-backed operator in the Eagle Ford, a small independent in the DJ Basin — these companies need smart engineers more than Chevron does. And they will give you more responsibility faster.
Here is why. A supermajor has structured programs, defined roles, and layers of hierarchy. You will do your assigned task and stay in your lane for two years. A small operator does not have that luxury. You will touch production, drilling, completions, and regulatory because there is nobody else to do it. You will learn more in one year at a 200-well company than in two years at a major.
These operators are also less likely to have sophisticated AI deployments, which means the grunt work still exists — and you can learn from it. Then, when you bring AI tools to bear on that work, you understand what the tools are actually doing because you have done it manually first.
Position yourself as the "AI-native" engineer.
Not an engineer who also dabbles in AI. Not a data scientist who read a reservoir engineering textbook. The person who genuinely bridges both worlds — who understands the Darcy flow equation AND can build an MCP server, who has run a nodal analysis AND can write a Claude prompt that generates one.
This combination is rare and getting rarer. PE enrollment has been declining for years. Fewer graduates, with each cycle, have the combination of domain depth and technical fluency the industry needs. If you can credibly position yourself in that intersection, you are not competing for the few remaining traditional entry-level slots. You are creating a category of one.
How to Get the Experience When Nobody Will Hire You
This is the question that matters most, and it is the one with the least satisfying answer. Because the honest truth is: the traditional system of "graduate, get hired, learn on the job" is not working for a significant percentage of new engineers right now. So you have to build your own path.
Here are the options I see, in rough order of accessibility.
SPE student competitions and paper contests. These are underrated. The PetroBowl, the student paper contest, the student chapter technical talks — they force you to work on real problems, present to real engineers, and build a network of people who might hire you. They are also one of the few places where you can demonstrate domain knowledge and communication skills simultaneously. If your student chapter is active, use it aggressively.
University programs with applied AI tracks. A growing number of PE departments are adding data analytics and machine learning to their curricula. These are not enough on their own — you need to go beyond the coursework — but they signal to employers that you have foundational AI/ML literacy alongside your PE degree. The programs I see doing this well include Texas A&M's data analytics and AI offerings, Colorado School of Mines, and OU. If your department does not offer these tracks, take them through the CS or statistics department.
Open-source petroleum engineering projects. I keep coming back to this because it is the most direct analog to real work experience. When you contribute to petro-mcp, you are writing code that interacts with real petroleum engineering data structures. When you contribute to lasio, you are working with the same LAS files that every petrophysicist in the industry uses. These contributions show up on your GitHub profile, and they demonstrate to a hiring manager that you can do the work — not just that you studied it.
Collaboration programs. This is where I will mention something we are doing at Groundwork Analytics, because I believe it is relevant and I want to be transparent about it. We built an open collaboration program specifically for this moment — for PE students and new graduates who want to work on real AI projects for the oil and gas industry, get mentorship, and build portfolio pieces that prove they can do the work. It is one option among several, and I am not pretending it is the only path. But if you are looking for a way to get hands-on experience with AI tools applied to real petroleum engineering problems, details are at petropt.com/collaborate. You can also explore our open tools to see the kind of work we do.
Self-directed projects with public data. This requires the most discipline but is available to everyone. Download production data from the Texas RRC. Pull completion data from FracFocus. Build a decline curve analysis pipeline in Python. Create type curves for the Wolfcamp. Publish your work. Write about what you found. The hiring manager who sees a candidate with a portfolio of self-directed analysis using real public data is going to pay attention — because it shows initiative, technical skill, and domain curiosity all at once.
The Uncomfortable Truth
I want to close with something that might sound contradictory but is not.
The bar IS higher. The training ground IS shrinking. The entry-level job market IS harder than it was five years ago, and AI agents are a significant part of why.
And yet.
Petroleum engineering is not going away. The world still runs on hydrocarbons. The Permian Basin is still producing more oil than most countries. Wells still need to be drilled, completed, and produced. Reservoirs still need to be managed. The physics does not care about AI hype cycles.
The Bureau of Labor Statistics projects petroleum engineering employment to grow 2.4% by 2033. The industry needs roughly 40,000 workers to replace retirees over the coming decade. The AI in oil and gas market is growing at 13% annually, creating demand for people who can bridge the domain and the technology. SLB, Halliburton, and Baker Hughes have all launched internal AI bootcamps because they cannot find enough engineers who understand both worlds.
The engineers who thrive will be the ones who combine domain knowledge with AI fluency. Who can look at a production surveillance dashboard generated by an AI agent and know — from field experience, from reservoir understanding, from pattern recognition built through real work — whether the agent's output makes sense or is dangerously wrong. Who can direct AI tools to solve problems that the tools themselves cannot identify.
That combination does not come from a classroom alone. It comes from building things. From analyzing real data. From contributing to real projects. From getting your hands dirty with code and completions and production data — even if nobody is paying you to do it yet.
The traditional ladder had rungs. Some of those rungs are being removed. But the distance you need to climb has not changed. The industry still needs you to get from "recent graduate" to "engineer who can make decisions." You just have to be more intentional about how you climb.
Start building now. Contribute to open-source PE tools. Analyze public datasets. Learn to use Claude Code, Cursor, and the AI tools that are reshaping the work. Understand the petroleum engineering deeply enough to direct those tools effectively and catch their mistakes. Target small operators who will give you real responsibility. Get field experience if you can. Join SPE. Present your work.
The bar is higher. But you can clear it. You just cannot wait for someone to hand you a stepladder.