Skilled Trades, Manufacturing & Energy

How to get hired as a CNC machinist in 2026-27

Updated 6 October 2026 · 53 min read · by SkillSculptor

The short answer

You get hired as a CNC machinist by passing a shop trial, not by writing a good application. The usual sequence is a short phone call, a floor interview with the shop manager or lead machinist, then a practical: read a print, measure a part, and set up a job at a machine while someone watches how you work. No US state licenses machinists, so no exam or license gates the job legally. What gates it is demonstrated setup ability, claimed on a one-page resume that names the exact machines, controls, materials and tolerances you have held, and then confirmed at the spindle, often within a week or two of applying.

License requiredNone. No US state licenses CNC machinists the way states license electricians, plumbers or barbers. There is no board exam, no continuing education requirement and no renewal. Anybody selling you a "CNC license" is selling you a school certificate. This cuts both ways: you can be hired on Monday with no paper at all if you can run the machine, and you cannot hide behind paper if you cannot.
The credential employers actually recognizeNIMS credentials from the National Institute for Metalworking Skills, awarded per skill area (measurement, materials and safety; job planning; CNC milling and CNC turning at operator, setup and programming levels) and earned by passing a written theory exam plus a performance test on a real part that gets inspected against a spec. NIMS is the closest thing the trade has to a common currency. It helps most for a career changer with no shop history, matters less once you have three years of parts behind you, and never replaces the trial.
Time to employableMonths, not years, for an operator or trainee seat. A community college machine tool technology certificate is commonly one to two semesters and an associate degree two years, both usually with NIMS credentials embedded. Many people skip school entirely, start as a machine operator loading parts, and move to setups over roughly one to three years, depending almost entirely on whether the shop lets them touch setups. Journeyman depth, meaning you can take a print and a block of stock to a finished part with no help, realistically takes four to six years at the machine.
The apprenticeship routeRegistered apprenticeships for Machinist, CNC Setup Programmer and Tool and Die Maker exist through employers, the IAM and UAW where shops are organized, state apprenticeship agencies and manufacturing associations. A registered machinist apprenticeship commonly runs about four years and in the region of 8,000 hours of on the job training plus several hundred hours of related classroom instruction, with a scheduled wage ladder. Tool and die is usually longer. Search apprenticeship.gov and your state apprenticeship agency, and ask local shops directly: many run unregistered in-house programs that are just as good and never get advertised.
What the interview really isA practical. The common shape is a print reading and GD&T quiz, a measurement test (you are handed a part and some gauges and asked for real numbers), and a machine trial of 60 minutes to a full paid shift in which you indicate a vise, pick up a work offset, touch off tools, prove out a program and hold a dimension. Programming seats are tested by being sat in front of a CAM seat with a print, or by being asked to hand-edit a program at the control.
The gate nobody mentions in the postingUS person status. Aerospace and defense shops whose work falls under ITAR can only put US persons on controlled jobs, so a posting that looks open can be closed to a visa holder before the first call. Add a pre-employment drug screen (many defense and aerospace shops still test for THC regardless of state law, because federal contract terms drive it), a lift and vision requirement, steel-toe boots, and at some sites a background check or a clearance. Ask about citizenship requirements and the drug panel in the first conversation rather than after two interviews.
Site tickets that are not a license but will stop a start dateMany shops want an OSHA 10 or OSHA 30 general industry card, a forklift evaluation, overhead crane and rigging sign-off for anything over a few hundred pounds, lockout tagout training, and a respirator fit test if you will be near grinding or certain coolants or alloys. These are site requirements and training, not state licenses, and most are provided by the employer in the first week. Having the OSHA card already is a cheap, visible differentiator on an entry-level resume.
Where to get real pay numbersBLS Occupational Employment and Wage Statistics, by SOC code, read at your own metro and your own percentile rather than nationally: 51-4041 Machinists, 51-4111 Tool and Die Makers, 51-9161 Computer Numerically Controlled Tool Operators and 51-9162 Computer Numerically Controlled Tool Programmers. Then cross-check against live postings in states with pay transparency rules that require a range in the advert, and against union scale if your area has organized shops. The published Machinists category mixes manual with CNC and operators with setup hands, so its median sits below what a setup and programming machinist is paid.
Shift realityMost shops run two or three shifts. Second shift (roughly 3pm to 11:30pm) and third shift (roughly 11pm to 7am) carry a per-hour differential and have the most openings, the least competition and often the fastest promotion, because there are fewer people on the floor and you get handed your own setups sooner. Overtime is common and in many shops is a large part of what the year actually pays. Four tens and three twelves are both normal. Ask what the real weekly hours have been for the last six months, not what the posting says.

How CNC machinist hiring actually works

Forget the four-stage corporate loop. A large share of CNC machinists in the US work in small and mid-size job shops and contract manufacturers, places with ten to two hundred people, where the person who decides is the shop manager, the production lead or, in a small shop, the owner. There may be an HR coordinator who schedules you and runs the drug screen, but they are not evaluating your machining. In a busy market the whole process runs in a week: apply Monday, floor interview Wednesday, trial Thursday, offer Friday, start the following Monday.

The order is usually a ten minute phone call that checks three things (do you really have the machine time you claim, which shift can you work, and can you pass a drug screen), then an on-site interview that is half conversation and half walking the floor, then a practical. The floor walk is not a courtesy tour. They are watching whether you put on safety glasses without being told, whether you look at the machines and the parts, and whether you ask a real question. Candidates who stare at their phone in the shop do not get the trial.

Larger employers run a longer process. Aerospace and defense primes, medical device manufacturers and semiconductor equipment suppliers have real HR functions, applicant tracking systems, structured interviews, background checks and sometimes a multi-week export control review before a start date. Expect two to six weeks there instead of one, and expect the resume itself to matter more because a keyword filter sits in front of a human. The trade-off is worth understanding: a job shop will judge you almost entirely on what you can do at the machine, while a large manufacturer will judge your paperwork first and your hands second.

There are five doors in, and they do not open the same way. Direct to the shop, by applying or by walking in with a folder. Through a staffing agency on a temp-to-hire basis, which is the fastest door and the lowest-paid one, and which the shops hiring in volume use constantly. Through a school, where the instructor's phone is often the actual placement mechanism and employers call the program before they post anything. Through an apprenticeship, registered or in-house. And through a referral, which in this trade includes a channel outsiders never think of: the tooling distributor rep who sells endmills and inserts to every shop in your county knows exactly who is hiring and who is losing people, and so does the machine tool dealer's service technician.

A note on job boards. Postings undersell this market badly. Many shops never post, because the last three people they hired came from a referral or a walk-in and a posting generates a hundred unqualified applications they have nobody to screen. If you are serious about a geography, build a list of every machine shop within your commute from a business directory or a drive around the industrial parks, and contact them in order. A mid-size metro will give you dozens of names, and that list will outperform any amount of time spent on a job board.

Nothing licenses you: what actually qualifies a machinist

This is the single most useful fact for anyone entering the trade, and it is routinely misunderstood: there is no machinist license in the United States. No state board, no exam, no renewal, no reciprocity. A machine shop can hire anyone it wants to stand at a machine tomorrow. That makes machining one of the most open skilled trades to enter and one of the hardest to bluff your way through, because when the paper means nothing the test means everything.

What does exist is a credential system that signals competence to people who cannot yet watch you work. NIMS, the National Institute for Metalworking Skills, publishes skill standards and awards credentials per area: measurement, materials and safety; job planning, benchwork and layout; CNC milling and CNC turning at operator, setup and programming levels; grinding; EDM; and more. A NIMS credential requires both a written theory exam and a performance test where you actually make a part that is then inspected against the standard. That performance component is why employers take it more seriously than a classroom certificate. Where it helps most is at the start: a career changer holding a NIMS CNC Milling Setup credential has something checkable, where a career changer with nothing has only a claim.

Education sits alongside, not above. A community college machine tool technology certificate (commonly one to two semesters) or an associate degree (two years) will teach you print reading, GD&T, shop math, manual machining, CNC setup and a CAM package, usually with NIMS credentials built into the coursework and often with an internship attached. It is a genuinely good route, especially if the program has current machines and real industry contacts. It is not a requirement, and a certificate from a program with three worn-out mills and no employer relationships is worth very little. Visit before you enroll, look at the machines, ask which control they teach on, and ask for the placement rate and the names of the shops that hired last year's class.

Then there are the gates that sit around the job rather than on it. Export control is the big one. Work controlled under ITAR requires that the person touching it is a US person, which covers US citizens, lawful permanent residents and certain protected individuals such as asylees and refugees, and which closes a large share of aerospace and defense shops to visa holders before any skill question arises. Quality systems bring their own requirements: AS9100 aerospace shops, ISO 13485 medical device shops and IATF 16949 automotive shops all run documented processes, first article inspection to AS9102, lot traceability and controlled revisions, which means they want someone who writes things down, signs for what they did, and does not change a process without routing it. Nadcap applies to special processes such as heat treat and plating that often sit downstream of your parts.

Finally, the quiet screens. A pre-employment drug test is standard, and many aerospace and defense shops still screen for THC regardless of what state law says about recreational use, because their contracts and their customers require it. Steel-toe boots and safety glasses are yours to bring, on day one and usually to the interview. Some shops require a lift test or a basic physical, and color vision can matter for reading color-coded tooling and gauges. None of these are hard to meet, but all of them have cost people a start date they thought they had.

What gets you the job, after all that, is evidence of setup ability. Not years in a shop, not a certificate, not how many machines you can name. Evidence that you can take a print and a piece of stock, decide how to hold it, pick up your offsets, prove out the program without crashing, measure the result honestly, and make it again tomorrow.

Operator, setup, programmer: know which job you are applying for

"CNC machinist" covers at least four different jobs with different pay, different hiring tests and different futures. Applying without knowing which one the posting means is how good candidates get screened out for being either over or under qualified, and it is how people end up accepting an operator wage for setup work.

The CNC operator loads and unloads parts, starts cycles, deburrs, gauges to a control plan and adjusts an offset when a dimension drifts. The setup is already made. This is the entry seat, it is learnable in weeks, and in a production shop it can be repetitive. It is also the single most common way people get into the trade, and there is no shame in it. The question to ask in the interview is not whether the job is operating. It is what the documented path from operating to setup looks like, and how long it took the last two people who walked it.

The CNC setup machinist is the job most postings mean when they say machinist. You take a job packet, pick workholding, build or pull a fixture, load and touch off tools, pick up work offsets, prove out the program, run first articles, get them inspected, dial in the offsets and hand the job to an operator or run it yourself. You own the first good part. This is where the trade's real skill sits and where the pay step is largest, because a shop's throughput is limited by setup time and a hand who can set up a job in forty minutes instead of two hours is worth a great deal of money.

The CNC programmer or programming machinist takes a model and a print, decides the process (how many operations, which machine, what order, where the datums go), builds toolpaths in CAM, posts the code, simulates it, creates the setup sheet and tool list, and often proves it out at the machine. In a small shop this is the same person as the setup machinist. In a larger shop it is a separate seat, sometimes called manufacturing engineer or NC programmer, sometimes sitting in an office. Pay is usually higher and the work is less physical, which matters more at forty-five than at twenty-five.

Above and beside those sit the specialists and the leads: toolmaker and tool and die maker (dies, molds, fixtures and gauges, usually the longest training in the trade), Swiss machinist (small diameter turned parts on a sliding headstock lathe, a genuinely separate skill that is in chronic short supply and pays accordingly), grinder hand, EDM operator, CMM programmer, applications engineer at a machine tool or cutting tool supplier, and shop supervisor. Several of those are better endpoints than shop supervisor, which is often a pay cut per hour once you lose overtime.

Read postings for which seat they mean, not for the title. "Must be able to set up and operate" means setup. "Load and unload, inspect to control plan" means operator. "Proficient in Mastercam, create programs from solid models" means programmer. If the posting says machinist and the duties say operator, the wage will say operator too.

The routes in, and what each one really costs

Nobody needs all of these. You need the one that fits your money, your time and your local market. The costs below are the honest ones: time, wage sacrificed, and what you hold at the end.

Pick the route by asking what your local shops actually hire from. In a region with a strong community college program, the certificate route is fast and well connected. In a defense or aerospace corridor, apprenticeships and in-house training programs are funded and plentiful. In a market full of small job shops, walking in and starting as an operator may put you at a machine a year earlier than any school would.

The shop trial: how to pass the part of hiring that decides it

Almost every serious machinist job ends with a practical, and the practical is the decision. The forms it takes, roughly in order of how common they are: a print reading and GD&T test on paper, a measurement test where you are handed a part and some gauges and asked for numbers, a shop math and speeds and feeds test, a machine trial where you set up and run something, and for programming seats a timed CAM exercise or a hand-edit at the control. A full paid shift as a working trial is also common and is usually a good sign, because a shop that pays you for a day intends to hire someone.

Ask what the trial will involve before you arrive. Most shops will tell you, because they want you to pass. If they will not say, assume print reading, measurement, and a vise setup on a mill or a chuck setup on a lathe.

Bring your own safety glasses (prescription if you need them), steel-toe boots, a six inch caliper and a 0-1 inch micrometer if you own them, a dial test indicator, a six inch scale, a deburring tool, a pen and a notebook, and copies of any credentials. Owning a basic tool roll signals that you are a machinist rather than a person who has been near machines. Some shops supply everything, and none of them mind that you came prepared.

The single most important behavior in a trial is narrating your process. The evaluator cannot see you think. Say what you are doing and why: "the print has a position callout on the bore at MMC referenced to A, B and C, so I want to pick up my X and Y from the datum faces and not the stock edge", "I am going to indicate the vise in to under half a thou before I touch anything", "I will prove this out in single block with rapid at twenty five percent and my hand on feed hold". A silent candidate who gets the right answer reads as lucky. A narrating candidate who hits a problem and reasons through it reads as safe to leave alone, which is the actual thing being assessed.

The measurement test is where more candidates fail than anywhere else, and the failure is almost always the same one: using the wrong instrument for the tolerance. If a bore is 0.7500 plus zero minus 0.0005, calipers cannot tell you anything useful. The answer is a bore gauge mastered against a setting ring or a micrometer, or gauge pins, and knowing which you are using and why. State your instrument, state its resolution, deburr and wipe the part before you measure, let a hot part come down to temperature, and take the reading more than once. Mention the ten to one rule (your gauge should resolve roughly ten times finer than the tolerance you are judging) and you will be ahead of most of the room.

The machine trial itself is a sequence, and doing it in order is most of the grade. Read the whole print first and say what the critical features and datums are. Plan the operations out loud: how you will hold it, in what order, what gets left for a second op, where the stock allowance goes. Indicate the vise or chuck. Pick up the work offset with an edge finder, a dial indicator or a probe, then verify it by jogging to a known feature before you cut. Load tools and touch off lengths. Prove out above the part: single block, rapid override down, distance to go on the screen, finger on feed hold. Run the first part. Deburr it. Measure it. Adjust the offsets. Run the second. Then tell them what you would change for production: a soft jaw instead of a parallel, a different tool to lose a tool change, a tool life setting, a cycle time estimate.

What they are grading, beyond the part: did you wear safety glasses without being asked, did you take gloves off before touching a running lathe, did you put tools back, did you sweep the chips and wipe the table down, did you ask when you did not know instead of guessing and crashing, and did you tell the truth about a dimension that was out. Shops will forgive a candidate who misses a dimension and says so. They will not hire one who shows them a part that is out of tolerance and calls it good, because that is the single behavior that costs a shop a customer.

For a programming trial, expect a print and a seat of whatever CAM the shop runs, with sixty to a hundred and twenty minutes to program a part. You are not being graded on finishing. You are being graded on process choice: operation order, datum strategy, workholding, tool selection, how you handle the feature that cannot be reached, and whether you simulate. If the shop runs a CAM package you do not know, say so immediately and ask to show your process in the one you do know. Transferable process thinking is the thing they want, and the button locations are a fortnight's work.

The resume a shop manager actually reads

A machinist resume is read for about twenty seconds by someone standing up, and they are looking for four things: which machines and controls, which materials, which tolerances, and whether you set up or only ran. Everything else is noise. One page, no photo, no objective statement, no soft-skill adjectives. If you have twenty years, it is still one page plus an optional second page that is purely a machine and capability list.

Name machines with models and controls, not categories. "CNC mills and lathes" tells a shop nothing. "Haas VF-2SS and VF-4 (Haas control), Haas UMC-750 5-axis with Renishaw probing, Mazak QTN 250MSY mill-turn (Mazatrol SmoothG), Doosan Puma 2600 (Fanuc 0i-TF), Citizen L20 Swiss (Fanuc)" tells them exactly whether you can walk up to their machine on Monday. The control matters as much as the machine: a Fanuc hand and a Mazatrol hand are not interchangeable on day one, and a shop that runs Heidenhain or Okuma OSP is scanning for exactly that word.

Name tolerances you actually held, with the feature attached. "Tight tolerances" is a phrase with no information. "Held bore diameters to plus or minus 0.0002 in with true position 0.0005 at MMC, 16 microinch Ra on sealing faces, in 17-4 PH stainless" is information, and it is also checkable in the trial, which is why it reads as credible. Do the same with materials (6061 and 7075 aluminum, 303 and 316 stainless, 17-4 PH, 4140 pre-hard, Inconel 718, Ti-6Al-4V, PEEK, Delrin, hardened tool steel) and with lot sizes and part sizes, because a prototype shop and a hundred thousand piece production shop are looking for opposite things.

Say explicitly what you owned. Setup from a job packet? Process planning from a model? Program creation in CAM, or editing at the control? First article inspection? CMM programming? Fixture design? Tool ordering? Training operators? Those phrases are the ones a hiring manager scans for, because they separate a setup machinist from an operator, and candidates lose the setup wage by writing the operator version of their own job.

Quantify the things that are genuinely measurable in a shop, using your own real numbers: setup time reduced on a repeat job, cycle time cut and how (tooling change, different approach, dropping an operation), scrap reduced, machines run simultaneously, parts per shift, first-pass yield on first articles. A line in the shape of "reduced setup on a recurring 300-piece job from 2.5 hours to 50 minutes by building a soft jaw fixture and standardizing the tool list" is the most persuasive thing a machinist resume can contain, because setup time is the constraint in nearly every job shop. Write the version that is true of you, and be ready to be asked how.

List quality systems and inspection experience plainly: AS9100, ISO 9001, ISO 13485, IATF 16949, ITAR registered facility, first article inspection to AS9102, SPC, Gage R&R, CMM programming in PC-DMIS or Calypso, Keyence vision gauges, optical comparator, surface plate work. For aerospace and medical shops this block decides whether you are interviewed at all, because their own customers audit it.

What gets ignored or actively hurts: an objective statement, "hard worker" and "team player", a skills bar chart, GPA, high school details if you have shop experience, and any claim of 5-axis or programming that you cannot demonstrate in twenty minutes. List unrelated past jobs one line each for continuity, because unexplained gaps cost you more than irrelevance does. Over-claiming is the most expensive error on a machinist resume, because the trial is right there.

Bring a portfolio. Photographs of parts you made, on your phone or printed, with a line each: material, what was hard about it, the tolerance, the machine. Respect the rules: never photograph a controlled print or an ITAR part, never take a customer's drawing out of a shop, and if you are not sure, describe the part in words instead. A machinist who shows five parts and explains the fixture on the awkward one has effectively passed half the interview before the trial starts.

What the interview actually tests, and the answers behind the questions

The conversation around the trial is shorter than in office hiring and more technical. There are no behavioral frameworks and nobody wants to hear about your greatest weakness. There are roughly six things being checked, and knowing which is which lets you answer the real question.

Print reading and GD&T. Expect to be handed a drawing and asked what a callout means. Be solid on datums and datum reference frames, basic dimensions, position with maximum material condition and the bonus tolerance that comes with it, profile of a surface, flatness, perpendicularity, runout and total runout, surface finish callouts in Ra, and title block conventions including the default tolerance and the revision level. Know which edition of ASME Y14.5 the drawing invokes, because it is printed in the title block and it changes things: the 2018 edition dropped concentricity and symmetry as symbols, so a print still carrying them is drawn to an older edition. If the shop uses model-based definition, with the tolerances carried as PMI inside the 3D model rather than on a 2D sheet, say whether you have worked that way, because it is spreading in aerospace and automotive and most candidates have not.

Shop math and speeds and feeds. The common ones: convert surface feet per minute to spindle speed (RPM is roughly 3.82 times SFM divided by the cutter diameter in inches), compute feed rate from chip load, flutes and RPM, find bolt circle coordinates, use a sine bar to set an angle, pick a tap drill size, and convert a position deviation in X and Y into a true position diameter (twice the square root of the sum of the squares). Nobody minds a calculator. They mind not knowing which formula applies.

Troubleshooting, which is the heart of it. Expect open questions with no single right answer: the finish on this wall is poor, what do you look at; this bore is coming out oversize and out of round, what are the candidates; this part is moving in the vise; the tap keeps breaking in 304; this cut was fine for two hundred parts and now the dimension is drifting; the finish pass is chattering on a long thin feature. Good answers walk the system in order. Is it the part (fixturing, stock condition, stress relief), the tool (wear, runout, stickout, geometry, coating), the process (speeds, feeds, depth of cut, climb versus conventional, coolant, chip evacuation), or the machine (backlash, spindle, way wear, thermal growth). Naming the order you check in is worth more than guessing the single cause.

Scrap and honesty. Nearly every shop asks a version of "tell me about the worst part you scrapped, or your worst crash". There is only one wrong answer, which is that you never have. Tell a real one: what happened, how you found out, what you told your lead, what it cost, and the specific thing you changed afterwards. This question is not about competence. It is about whether you report a problem or hide it, and a shop that has been burned by a quiet machinist asks it first.

Reliability, which is quieter and more decisive than anyone admits. Shops lose more production to absence than to skill. Expect direct questions about attendance, transport, whether you can be there at 6am, whether you can work Saturdays, and why you left each previous job. Answer plainly and specifically. If you have a patchy history, name it, say what changed, and offer something verifiable. A candidate with mediocre skills who shows up every day beats a talented one who does not, and every shop manager has learned that the expensive way.

Safety and shop conduct. Partly tested in the trial, partly asked: gloves and loose clothing near rotating work, eye protection, chip handling with a hook and not a hand, lockout and tagout for maintenance, how you handle metalworking fluid and skin, hearing protection, and what you do when you are asked to run something you think is unsafe. The last one matters, and the answer is that you stop, say why, and ask, rather than either running it quietly or walking off.

Finally, your own questions. Ask them, and ask specific ones: what is the mix of prototype and production, what is the average lot size, how many setups a day does this seat do, who programs, what has the overtime pattern been over the last six months, what is the path from this seat to the next one, and who was the last person who moved up and how long did it take. Those questions mark you as somebody who intends to stay, which is the thing shops are short of.

Who is hiring, where the work is, and how to get a real pay number

CNC machining is not one market. The industries buying machining time in 2026 and 2027 have different cycles, different standards and very different day to day work, and choosing the sector is as consequential as choosing the shop.

Aerospace and defense is the strongest pull and the most restrictive. Defense production and the submarine industrial base in particular have been recruiting machinists hard across the supplier network, with public campaigns aimed directly at this trade (the Build Submarines recruiting effort is the most visible), and commercial aerospace build rates keep tier two and tier three suppliers busy. The work is often titanium, Inconel and tight tolerance aluminum, under AS9100 with full traceability and first article inspection, which means slow, documented and well paid. The catch is export control: a US person requirement on controlled work, plus background checks and sometimes a clearance.

Medical device is the other high-discipline sector. Implants, instruments and components under ISO 13485, a great deal of it small diameter Swiss turning in 316L, titanium and PEEK, with validation, lot traceability and surface finish requirements stricter than anything in general engineering. Swiss machinists are chronically short in this sector, which is the clearest skills arbitrage available to someone choosing what to learn next.

Semiconductor equipment and vacuum work has grown with fab construction, and the parts are large aluminum and stainless chambers and flanges with exacting flatness, finish and cleanliness requirements, often with cleanroom handling and ultra high purity specifications. It is less glamorous than aerospace and frequently pays comparably. Energy covers oil and gas downhole tooling and valves (regionally concentrated and cyclical with commodity prices), turbines, and a growing nuclear supply chain around small modular reactor components. Mold and die, including wire and sinker EDM and high speed graphite machining, is a separate craft with its own long training and its own shortage. Automotive and EV work is mixed, with prototype and tooling steadier than production. And underneath all of it sit general job shops, which employ more machinists than any single industry and will hand you more variety than any of them.

Geography is the biggest single determinant of your options and your pay. Machining clusters in the upper Midwest (Wisconsin, Michigan, Ohio, Illinois, Indiana), the Connecticut River valley and New England for aerospace, the Pacific Northwest, southern California, Arizona and Texas, the Carolinas and Tennessee, Florida's space coast, and the Gulf Coast for energy. Within a metro, machining lives in specific industrial parks, and a drive through one with a notebook is a better prospecting tool than any job board.

On pay, go to the sources rather than to any article's figures, this one included. BLS Occupational Employment and Wage Statistics publishes hourly and annual wages at the 10th, 25th, 50th, 75th and 90th percentile, nationally, by state and by metro area, for SOC 51-4041 (Machinists), 51-4111 (Tool and Die Makers), 51-9161 (Computer Numerically Controlled Tool Operators) and 51-9162 (Computer Numerically Controlled Tool Programmers). Read your own metro at the percentile you expect to occupy, not the national median. Be aware that the Machinists category mixes manual with CNC and operators with setup hands, which pulls the median below what an experienced setup and programming machinist is worth.

Then triangulate. A growing number of states require a pay range in job postings, so you can read live local ranges for the exact seat you want, and even if you do not live in one of those states you can see how a national employer prices the same job where disclosure is required. If your area has organized shops, published union scale is a hard number. And ask directly in the interview: "what is the range for this seat, and what did the last person hired into it start at" is a normal question in this trade and costs you nothing.

Finally, read the whole compensation rather than the hourly rate. The stack is base rate, shift differential for second and third, overtime (which in a busy shop is a large part of what the year pays, and which disappears when the shop is slow), production or attendance bonuses, tool allowance, boot allowance, tuition reimbursement, and the health plan and retirement match, which vary enormously between a fifteen person job shop and a prime contractor. Two offers a dollar apart per hour are often several thousand dollars a year apart in the other direction once overtime and benefits are counted.

Working with AI in this role

What AI has actually changed for CNC machinists, and what it has not

Start with the honest part, because machining is a favorite illustration for automation stories and most of those stories are wrong about where the work is. The core of this job is not being automated. The limiting step in a job shop is not cutting metal, it is setup: deciding how to hold an awkward part, building or modifying the fixture, indicating it in, picking up offsets, proving out without crashing, and fixing the three things that are not like the drawing. That work involves physical judgment, improvisation around a part that moved, and responsibility that attaches to a person. Nobody has a system that does it. A shop that quotes hundreds of part numbers a year and runs lots of twenty five cannot amortize automation across a job the way an engine plant can, and that describes most of the industry.

What has genuinely changed is everything around the spindle, and it has changed enough to matter in hiring. The real shift is lights-out and reduced-attendance running: pallet pools and pallet changers on machining centers, bar feeders and parts catchers on lathes, robot and cobot tending cells now cheap and usable enough for a shop with no robotics staff, and in-process probing that lets a machine measure, compensate and continue without a human. The economics are simple. The machine is the capital, and a shop that runs it sixteen or twenty hours a day instead of eight beats a shop that does not. The person who makes that possible is a machinist who can build a process reliable enough to leave alone, which is a different and more valuable skill than being fast at the machine.

Be skeptical about where the AI label gets applied. A large share of what is marketed as AI in CAM is not new and is not a language model. Adaptive and trochoidal toolpathing (Mastercam Dynamic Motion, SolidCAM iMachining, Fusion adaptive clearing), automatic feature recognition and rules-based automation have been in these packages for years and are algorithmic. They are extremely useful and you should know them cold. They are not the thing that appeared in the last two years. Saying so in an interview, rather than parroting a vendor line, reads as somebody who actually uses the software.

What is genuinely new and is reaching shop floors: machine monitoring that pulls live data off controls over MTConnect and OPC UA and turns it into utilization, cycle time variance and downtime analysis (MachineMetrics, Datanomix, Amper and similar platforms are now common in mid-size US shops); automated and AI-assisted quoting and RFQ processing, which is reshaping the estimating desk at both the platforms (Xometry, Protolabs, Fictiv) and the shops that use quoting software such as Paperless Parts; adaptive control that modulates feed from spindle load in real time and detects tool wear and breakage (Caron Engineering TMAC and the control vendors' own versions); automated inspection, from Keyence style vision gauges that take dozens of dimensions in seconds to CMM programs generated from a model carrying its tolerances as PMI; and general purpose language models used on the floor for exactly the tasks they are good at.

That last one deserves specifics, because it is the part a candidate can demonstrate. Useful on a shop floor today: explaining an unfamiliar block of G-code or an alarm code, drafting and debugging Fanuc custom macro B or parametric programs, generating a starting point for speeds and feeds that you then verify against the tool manufacturer's data, writing setup sheets and operator work instructions from your own notes, translating a work instruction for a shift where English is a second language, and summarizing a long specification into the handful of requirements that touch your operation. Dangerous on a shop floor today: running generated G-code. A model will confidently produce a program that rapids through a vise. Anything it writes goes through CAM or standalone simulation, then single block with rapid override down and a finger on feed hold, exactly like code from any other source. A machinist who says this out loud in an interview sounds like somebody who can be trusted with the tooling budget.

The practical summary for someone applying in 2026 and 2027: AI has not come for the setup, and claiming it has will make you sound like you read a headline instead of standing at a machine. Automation and monitoring have come for the attended run time, and the machinists who are gaining are the ones who can make a job run unattended, keep a cell fed, interpret the data coming off the control, and use software assistance without surrendering verification. Test it yourself before you believe anybody, including this page: pull thirty machinist postings in your own metro and count the words. The ones that pay are probing, automation, robot cell, lights-out, 5-axis, Swiss and CAM programming. Count how many times the word AI appears, and let that set how much of your interview you spend on it.

Making a job run unattended, and proving it is safe to leave

This is the highest-return skill in machining right now. The economics of a shop are driven by spindle hours, and the difference between a job that needs a person every ninety seconds and one that can run four hours alone is pure margin. Making that true is skilled work: tool life monitoring and sister tooling, in-process probing between parts, chip evacuation and coolant management that will not fail at 2am, spindle load alarm limits set tight enough to catch a broken tool and loose enough not to false-trip, part present and part off detection, and a check routine that confirms the twentieth part is still the first part. Shops pay for this specifically, because it converts a machine they already own into more capacity.

Show it: Name the job and the duration, with your real numbers. The shape that lands: "set up a 400-piece aluminum bracket job on a Haas VF-3 with a pallet changer to run six hours unattended overnight: sister tooling on the two roughers, Renishaw probing the bore every tenth part with automatic offset update, spindle load limits set from the proved cycle, a chip auger check on the setup sheet, and no overnight crash on that job since." Then be ready to explain exactly which failure you were guarding against with each measure.

Probing and in-process gauging

Spindle and table probes (Renishaw, Blum, Marposs) have moved from a luxury to a common option on new machining centers, and they do three distinct jobs: cutting setup time by picking up work offsets and part orientation automatically, in-process measurement with automatic offset compensation, and tool breakage and length checking. A machinist who can write and use probing cycles collapses the setup time that limits the shop, which is the metric a shop manager watches. Plenty of candidates have a probe on their machine and have never used it beyond a canned edge pick-up, so real fluency here stands out immediately.

Show it: Say what you probed and what it saved. "Used Renishaw Inspection Plus macros to pick up work offsets and part rotation on a 5-axis trunnion job, which took setup from over an hour of indicating to about twenty minutes." Mention tool setting and breakage detection separately if you have used it, and be ready to discuss probe calibration and what you do when the probe reading disagrees with the gauge.

Running and troubleshooting an automated cell: robot or cobot tending, pallet pools, bar feeders

Cobot tending cells have crossed the price and usability threshold where a job shop with no robotics engineer can buy one, and they are taught by hand-guiding and tapping a tablet rather than by writing robot code. Pallet pools on horizontals and bar feeders on lathes do the same job mechanically. Whoever owns the cell in a shop stops being interchangeable: part presentation, gripper and fixture design, cycle interlocks, light curtain and guarding, fault recovery, and the judgment about which parts belong in the cell and which go back to a hand. In practice shops push simple repeat work into automation and put experienced machinists on the complex jobs, which raises the value of both ends.

Show it: Name the platform and the part family. "Set up and ran a cobot tending cell on a Haas ST-20 for a repeat 2,000-piece stainless fitting: designed the stacked part nest, set the gripper and the pickup tolerance, wrote the M-code handshake, and recovered faults without calling the integrator." If you have not run one, say so plainly and describe which of your current jobs you would move into a cell first and why. That answer is worth almost as much as the experience.

CAM fluency including the adaptive toolpath strategies, plus verification

CAM is the main leverage a programmer has over cycle time and tool cost, and the adaptive and high-efficiency strategies (Mastercam Dynamic Motion, SolidCAM iMachining, Fusion adaptive clearing, hyperMILL and PowerMill for complex 5-axis) change the economics of hard materials in particular, trading depth of cut for radial engagement and transforming tool life in titanium, Inconel and hardened steel. Verification is the other half and is non-negotiable as parts get more expensive: machine simulation in CAM and standalone verification in Vericut or NCSIMUL catch the collisions a human eye cannot see on a 5-axis move.

Show it: Name the package, the post processor work you have done and the result in your own numbers. "Programmed in Mastercam with Dynamic Motion roughing on 17-4 PH, which multiplied tool life on the half inch rougher several times over and took roughly a third off the roughing cycle. Verified all 5-axis programs in Vericut against the machine model before proving out." Say clearly which packages you have used in production versus taught yourself, because that distinction will be tested.

Machine monitoring data: MTConnect, OPC UA and the shop dashboard

A growing share of mid-size shops have machine monitoring installed, pulling cycle, utilization, alarm and downtime data straight off the controls. Most of them are not getting much from it, because nobody on the floor engages with it. A machinist who tags downtime honestly, reads cycle time variance to spot a tool wearing or a chip nest forming, and brings a data-backed case for a tooling change or a fixture is doing the job the system was bought for. It also converts your improvements into evidence, which is useful at review time and on the next resume.

Show it: "Used MachineMetrics cycle time data to find a consistent swing on a repeat job that turned out to be the operator waiting on the air blast, resequenced the program and took seconds off every cycle across a 5,000 piece run." Put your actual seconds and percentage in. If your shop has no monitoring, the equivalent is a kept logbook of cycle and setup times, which demonstrates the same instinct.

Reading model-based definition: tolerances carried as PMI in the 3D model

Aerospace and automotive customers increasingly ship a model with the product and manufacturing information embedded rather than a dimensioned 2D drawing, which changes how you plan a job, how you build a CMM program and where the authority for a tolerance lives. Shops moving to MBD find that most of their experienced hands are drawing readers only. Being able to open a model, interrogate the PMI, understand the datum reference frame in 3D and build an inspection plan from it makes you the person who takes those jobs.

Show it: Name the viewer or CAD you used and the customer context. "Worked MBD jobs from an aerospace customer in NX and a 3D PDF viewer, planned datum pick-up from the model PMI, and built the CMM program from the model rather than from a print." Mention ASME Y14.41 if you have worked to it, and be straightforward if your experience is 2D drawings only.

Using a language model as a drafting tool, with verification discipline

Models are genuinely useful on a shop floor for macro writing, explaining unfamiliar code and alarms, first-pass speeds and feeds, setup sheets, work instructions, and summarizing a long customer specification. They are also capable of producing G-code that will crash a machine with complete confidence. The hiring-relevant skill is not prompting, it is the verification habit around it: treat output as a draft from an unknown source, simulate it, single block it, check it against the tool manufacturer's data. Shops have already been burned by the opposite behavior, so stating the discipline is as valuable as stating the capability.

Show it: Be concrete and show the guardrail in the same breath. "I use a model to draft Fanuc macro B routines and to write the operator work instruction from my setup notes, then I verify every macro by dry running above the part. I do not run generated G-code without simulating it first." Keep it off the headline of your resume. It is a tool you use, not your trade.

What a screen is looking for

These are the terms that a resume screen, human or automated, is matching against for this role. Use the ones that are true of you, in the words the posting uses.

Mistakes that cost people this job

Writing "operated CNC machines" instead of naming machines, controls, materials and tolerances.

Write the specific line: "Haas VF-4 and UMC-750 (Haas control), Doosan Puma 2600 (Fanuc 0i-TF), 17-4 PH and 6061, held bore diameters to plus or minus 0.0002 in and 16 microinch Ra on sealing faces, lots of 25 to 500." A shop manager reads a machinist resume for machines, controls, materials and tolerances in about twenty seconds. If those four are not there, nothing else on the page gets read.

Claiming you program or run 5-axis when you have edited offsets or run a 3+2 job somebody else set up.

Say exactly what you did, because the trial will find out within twenty minutes and an inflated claim ends the interview on integrity rather than on skill. "I prove out and edit at the control, I have not programmed from a model in CAM" and "I have run 3+2 on a UMC-750 with probing, I have not done simultaneous 5-axis" are both perfectly hireable statements, and they build the credibility that gets you taught the rest.

Reaching for calipers in the measurement test when the tolerance is half a thousandth.

Match the instrument to the tolerance and say why out loud. A 0.7500 plus zero minus 0.0005 bore wants a bore gauge mastered to a setting ring, or gauge pins, not calipers. State the ten to one rule (your gauge should resolve roughly ten times finer than the tolerance), deburr and wipe the part before measuring, let a hot part come down to temperature, and take the reading twice.

Showing up to a shop interview in office clothes with no boots and no safety glasses.

Dress for the floor: steel-toe boots, your own safety glasses, no loose sleeves, no rings, hair tied back, no earbuds. Many shops will not take you past the yellow line without boots and glasses, so you lose the floor walk and the trial before anything technical happens. Bring a folder with the resume, your credentials and photographs of parts you made.

Wearing gloves near a running lathe, or any rotating workpiece.

Take them off before the spindle turns, every time. This is the fastest way to be walked off a trial in any shop with a safety culture, and it will not be explained to you afterwards. Same category of instant disqualifier: reaching into a machine past the door interlock, clearing chips with your hand instead of a hook, and leaning over a running chuck.

Proving out a new program at 100 percent rapid with the door shut and no hand on the controls.

Single block on, rapid override at about twenty five percent, Z retract checked, distance-to-go on the screen, and a finger on feed hold through the first approach of every tool. Then verify your work offset by jogging to a known feature before anything cuts. Evaluators watch for this sequence specifically, and it is the clearest single indicator of whether you can be left alone on a night shift.

Handing over a part that is out of tolerance and calling it good, or quietly fixing it later.

Say the number you measured, not the number you wanted. "This bore came in 0.0007 over, here is what I would change" is a hireable answer. Hiding it is not, and the "tell me about your worst scrap or crash" question exists precisely to find out which you are. Give a real example: what happened, how you reported it, what it cost, and the specific thing you changed afterwards.

Only applying to day shift, and only through job boards.

Apply for second and third shift deliberately: fewer applicants, a shift differential, and usually faster access to your own setups because there is nobody else to do them. And supplement the boards, because many shops never post. Build a list of every machine shop within your commute from a business directory, walk in mid-morning with a folder, register with two industrial staffing agencies, and ask the local cutting tool distributor rep who is hiring.

Taking an operator seat on a verbal promise of setup training with no definition of what triggers it.

Before accepting, ask three concrete questions: what is the progression from this seat, who was the last person who moved up and how long did it take, and what specifically has to be true for me to be doing my own setups. Get the answer in the offer conversation. Shops that cannot answer tend to be shops where you load parts for three years, and that is time you do not get back.

Assuming a posting is open to you when the shop runs ITAR controlled work.

Ask in the first phone call whether the role requires US person status or a clearance. Aerospace and defense shops are legally limited in who can touch controlled work, so the restriction is not negotiable and is not a judgment about you. Finding out on call one saves two interviews and a day off work, and it redirects you toward the medical device, semiconductor, energy and general job shop sectors where the gate does not exist.

Repeating vendor talk about AI running the machines, or the opposite, dismissing automation entirely.

Say the accurate thing: the setup is not automated and the attended run time increasingly is. Then point at what you can actually do with it. Probing, lights-out capable processes, robot or pallet cells, machine monitoring data, and your verification discipline for anything a model generates. "I use a model to draft macros and work instructions, and I never run generated G-code without simulating and single-blocking it" is the sentence that reads as competent in 2026.

Negotiating on the hourly rate alone.

Price the whole stack: base, shift differential, the real overtime pattern over the last six months (not the posting's claim), bonuses, tool and boot allowance, tuition reimbursement, health plan cost and retirement match. A job a dollar an hour lower with ten reliable overtime hours a week and a funded apprenticeship can be worth more than one a dollar higher that is slow. Ask what the last person hired into this seat started at. In this trade that question is normal.

Questions people ask

Do you need a license or certification to be a CNC machinist?

No. No US state licenses CNC machinists the way states license electricians, plumbers or barbers, and there is no board exam or renewal. Any program selling a "CNC license" is selling a school certificate. The credential employers do recognize is NIMS, from the National Institute for Metalworking Skills, awarded per skill area such as CNC Milling Setup or Measurement, Materials and Safety, and earned by passing both a written exam and a performance test on a part that is inspected against a standard. NIMS helps most for a career changer with no shop history and matters less once you have years of parts behind you. What actually gates the job is the shop trial.

How long does it take to become a CNC machinist?

Months to get hired, years to become good. An entry operator seat needs no credential and can be had in days. A community college machine tool certificate is commonly one to two semesters and an associate degree two years, both usually with NIMS credentials included. A registered machinist apprenticeship runs about four years and in the region of 8,000 hours of on the job training plus classroom instruction. Realistically, being able to take a print and a block of stock to a finished part with no help takes four to six years at the machine, and tool and die takes longer.

Can I get a CNC machinist job with no experience?

Yes, as an operator, which is how a large share of CNC machinists start. Shops hire people with no background to load and unload parts, deburr, gauge to a control plan and adjust an offset, then train the ones who show up every day and pay attention. The risk is being parked in that seat, so before accepting, ask what the documented path to setup is and who moved up most recently. Three things make you dramatically more hireable with no experience: your own boots and safety glasses, any NIMS credential or a completed free curriculum such as Titans of CNC Academy, and a willingness to work second or third shift.

What is a CNC machinist shop trial and how do I pass it?

A CNC machinist shop trial is a practical test that usually decides the hire. The common shape is a print reading and GD&T quiz, a measurement test where you are handed a part and gauges and asked for real numbers, and a machine trial of 60 minutes to a full paid shift in which you indicate a vise or chuck, pick up a work offset, touch off tools, prove out a program and hold a dimension. Pass it by narrating your process out loud, matching the gauge to the tolerance, verifying your offset before cutting, proving out in single block with rapid override low, deburring before you measure, reporting the real number including when it is out, and cleaning the machine afterwards. The evaluator is deciding one thing: can you be left alone.

Do I need to know CAM programming to get hired as a machinist?

Not for an operator seat and not for many setup seats, where editing at the control and running from a job packet is enough. You do need it for a programmer or programming machinist seat, and knowing it raises your pay everywhere. The package to learn depends on your market: Mastercam is the most common in US job shops, Fusion is cheap or free for individuals and widely used by small shops, Esprit is strong in Swiss and mill-turn, and SolidCAM, GibbsCAM, hyperMILL, NX CAM and PowerMill all have their sectors. The transferable thing is process planning, meaning operation order, datum strategy, workholding and tool selection. Button locations in a new package are a fortnight's work, and process thinking is years.

Is CNC machining being replaced by AI or robots?

Not at the core of the job. The limiting step in a job shop is setup: deciding how to hold an awkward part, building the fixture, indicating it in, proving out and fixing the things that are not like the drawing. That is physical judgment with liability attached, and no system does it. What has genuinely changed is the attended run time. Pallet pools, bar feeders, robot and cobot tending cells, in-process probing and adaptive control let machines run with far less human presence, and monitoring platforms pull live data off the controls over MTConnect and OPC UA. The machinists gaining from this are the ones who can make a job reliable enough to run unattended. Much of what is branded AI in CAM is in fact adaptive toolpathing and feature recognition that predates the current wave.

How much do CNC machinists make?

Go to the source rather than to any quoted band. BLS Occupational Employment and Wage Statistics publishes hourly and annual wages at the 10th, 25th, 50th, 75th and 90th percentile, nationally and by state and metro area, for SOC 51-4041 Machinists, 51-4111 Tool and Die Makers, 51-9161 CNC Tool Operators and 51-9162 CNC Tool Programmers. Read your own metro at the percentile you expect to occupy, and note that the Machinists category mixes manual with CNC and operators with setup hands, which pulls the median below what an experienced setup and programming machinist earns. Cross-check against live postings in states that require a pay range in the advert, and against union scale where shops are organized. Then price the whole stack: base, shift differential, overtime, bonuses and benefits.

Which industries hire the most CNC machinists?

General job shops and contract manufacturers employ the most CNC machinists, and give the widest variety of work. The sectors with the strongest pull going into 2026 and 2027 are aerospace and defense, including the submarine and munitions supply base, which pays well and runs AS9100 documentation but restricts controlled work to US persons; medical device under ISO 13485, much of it small diameter Swiss turning and chronically short of Swiss hands; semiconductor equipment and vacuum components, with large precision aluminum and stainless chambers; energy, covering oil and gas tooling, turbines and a growing nuclear supply chain; and mold and die, which is its own craft with its own shortage. Geographically, machining clusters in the upper Midwest, New England, the Pacific Northwest, southern California, Arizona, Texas, the Carolinas, Tennessee and the Gulf Coast.

Do I have to be a US citizen to work as a CNC machinist?

Not to work as a CNC machinist in general, but yes for a large part of the trade. Work controlled under ITAR can only be performed by a US person, which covers US citizens, lawful permanent residents and certain protected individuals such as asylees and refugees. That closes much of aerospace and defense machining to visa holders regardless of skill, and some sites add a background check or a security clearance. This is a legal restriction rather than a preference, so ask about it on the first phone call and do not take it personally. Medical device, semiconductor, energy and general job shop work is largely open, and those sectors are hiring.

What should go on a CNC machinist resume, and what gets ignored?

Put on a CNC machinist resume: machines with model and control ("Haas VF-2SS, Haas control; Mazak QTN 250MSY, Mazatrol Smooth; Citizen L20, Fanuc"), materials, the tolerances and finishes you actually held with the feature they were on, lot sizes, whether you set up or only ran, CAM packages, inspection and CMM experience, quality systems (AS9100, ISO 13485, AS9102 first article), your shift availability, and credentials by exact name. The single most persuasive line is a setup or cycle time improvement with your real before and after numbers. Ignored or harmful: objective statements, "hard worker" and "team player", skills bar charts, GPA, a photo, and any 5-axis or programming claim you cannot demonstrate in twenty minutes, because the trial is right there. One page.

What safety cards or site requirements do CNC machinist jobs ask for?

None of the site requirements a CNC machinist job lists is a state license, but several can hold up a start date. Common ones are an OSHA 10 or OSHA 30 general industry card, a forklift evaluation, overhead crane and rigging sign-off, lockout tagout training, and a respirator fit test where grinding or certain coolants and alloys are involved. Most employers provide this training in the first week, and most ask you to supply your own steel-toe boots and safety glasses from day one. Holding an OSHA 10 card already is cheap and visible on an entry-level resume. Separately, expect a pre-employment drug screen, and be aware that many aerospace and defense shops still test for THC regardless of state law.

Put this on a resume in about a minute

Paste your history once and point it at the CNC Machinist posting you are looking at. No account, no card.

Build my resume free More roles

Related roles