Choosing a field when money decides the path

Why money is a legitimate primary filter, not a shameful one

A lot of career advice pretends money is a secondary concern. Follow your passion, the story goes, and the income will sort itself out. That advice is usually written by people whose income has already sorted itself out. If you are 22 and staring at a student loan repayment schedule, or 34 and looking at a mortgage plus daycare, or 45 and reskilling because your prior industry collapsed, then the salary a field pays over the next five years is not a vulgar detail. It is the constraint that decides whether the plan is realistic at all.

This article is written for people who have decided, explicitly, that compensation is the primary axis of the decision. Not the only axis: nobody should walk into a job they will hate every morning for the next decade just because it pays well. But the primary axis. You want to know which fields in tech actually pay, how much runway it takes to get in, what the risk profile looks like, and how to translate a target salary into a concrete study plan.

We are going to be honest about tradeoffs. Some high-paying fields have brutal on-call rotations. Some have hiring cycles that punish anyone without a specific credential. Some pay a lot at the median but very little at the entry level, which matters if you cannot afford a two-year unpaid runway. Others pay less at the top but offer stable, predictable ladders that compound well over 15 years.

The framing is not "which field pays the most" in the abstract. It is: given your current savings, your current skills, your risk tolerance, and how long you can realistically study before you need income, which field gives you the best expected value? That is a different question, and it produces different answers for different people. A 19-year-old with parental support and four free years can rationally chase machine learning research. A 38-year-old career-changer with two kids probably cannot.

We will walk through the money-relevant fields in tech, the salary bands you should actually expect (not the outliers the recruiter posts on LinkedIn), the entry cost in months and dollars, the risk of the ladder collapsing under you, and how to build a study plan that respects your financial constraints instead of ignoring them. Refonte Learning built its career orientation mentor service around exactly this kind of conversation, because we found that generic advice was actively harming people who needed money-first plans.

The 2026 salary landscape: what the numbers actually look like

Before picking a field, you need calibrated expectations. Compensation in tech is bimodal and geographically warped, and public averages hide both effects. A "software engineer" in the United States can mean 75,000 dollars in Kansas City or 340,000 dollars in San Francisco at a FAANG-tier company, and both are the same job title. In France, the same title spans roughly 38,000 to 95,000 euros base. In India, from 6 lakhs to 60 lakhs. Any career decision made against a single number is a career decision made against a lie.

Here is a rough 2026 map for total compensation (base plus bonus plus equity vested annually) at the mid-career level, three to seven years of experience, in a major tech market:

• General software engineering: 110k to 200k USD, higher at top-tier firms
• Data engineering: 120k to 210k USD, premium for streaming and lakehouse skills
• Machine learning engineering: 150k to 320k USD, wide spread by employer tier
• AI research (applied): 180k to 500k USD plus, extremely employer-dependent
• Cloud security engineering: 140k to 240k USD, strong upward trend
• Site reliability engineering: 140k to 260k USD, on-call premium included
• DevOps engineering: 110k to 190k USD, overlaps with SRE at the top
• Data analytics: 80k to 140k USD, plateau earlier than data engineering
• Data science (product-facing): 120k to 210k USD, ambiguous ladder
• Full-stack web engineering: 100k to 180k USD, huge variance by employer
• Mobile engineering (iOS/Android): 115k to 200k USD, stable
• Solutions architecture (cloud vendor or partner): 150k to 280k USD, sales-adjacent bonus structure

These are working numbers, not gospel. They will be lower in most of Europe outside the UK, Switzerland, and the Nordics, and much lower in most of Asia and Latin America. But the relative ordering holds surprisingly well across geographies. AI research pays more than analytics almost everywhere. SRE pays more than junior full-stack almost everywhere.

The more important observation: the entry-level version of each of these pays substantially less than mid-career, sometimes by a factor of two to three. A junior data engineer at 65k is on a curve that reaches 180k in five years. A junior support engineer at 55k is on a curve that reaches 95k in five years. Both curves are real. Which one you can access depends on what you can plausibly study for and interview into. Our post on choosing by market demand breaks down which of these ladders currently have the strongest hiring signal.

Entry cost: the number nobody puts on the salary chart

Every high-paying field has an entry cost measured in months of study, tuition or subscription fees, and lost income. The salary chart above is meaningless without this second chart. A field that pays 250k after five years of unpaid graduate study is not, on a net-present-value basis, obviously better than a field that pays 140k after four months of bootcamp.

Rough entry-cost bands to reach a first paying job, assuming you start from moderate technical literacy (comfortable with a laptop, some scripting exposure):

• Full-stack web: 6 to 12 months of self-study or bootcamp, 0 to 20k in course costs
• Data analytics: 4 to 9 months, 0 to 8k in course costs, portfolio-driven
• Data engineering: 9 to 18 months, needs SQL, Python, and one cloud data stack
• DevOps: 9 to 15 months, needs Linux fluency, Terraform, Kubernetes
• Cloud security: 12 to 24 months, usually needs prior IT or dev experience plus certs
• SRE: 12 to 24 months, typically requires prior software engineering baseline
• Machine learning engineering: 18 to 36 months, math and systems both required
• AI research: 4 to 7 years including a graduate degree, huge opportunity cost
• Mobile: 8 to 14 months, platform-specific tooling learning curve
• Solutions architecture: rarely entry-level, usually 5+ years prior experience

Notice the pattern: the fields at the top of the salary chart are also at the top of the entry-cost chart. This is not a coincidence, it is the market pricing scarcity. If AI research were easy to enter, it would not pay 400k.

The practical implication for money-driven career choice is that you must optimize across both charts jointly. The field with the highest peak salary is not always the best expected-value bet. If you have six months of runway, machine learning engineering is not on your menu no matter how much it pays. Data analytics or a focused full-stack track is. If you have three years of runway and a math background, the calculation flips.

This is the specific conversation Refonte Learning mentors have with career-changers. It is not "what do you want to do," it is "what can you afford to study for, and what does that study realistically unlock?"

The high-ceiling fields: AI engineering and ML

Machine learning engineering and applied AI are the highest-ceiling fields in tech in 2026, full stop. Total compensation packages at the top of the market routinely exceed 500k USD for senior individual contributors, and specialist roles at frontier labs can go materially higher. If your money goal is aggressive, this is where the ceiling lives.

But the entry cost is real and the field is unforgiving of shallow preparation. To do the work, not just to hold the title, you need working fluency in linear algebra, probability, and optimization; comfortable Python and PyTorch or JAX; enough systems engineering to move data at scale; and either research chops (papers, novel work) or applied depth (shipping a model into production, monitoring it, iterating). The set of people who have all of that is smaller than the set of people who have taken a Coursera course on neural nets, which is why the pay is what it is.

The field also stratifies aggressively. There is a wide gulf between an "ML engineer" who fine-tunes off-the-shelf models with LoRA adapters and an ML engineer who designs training infrastructure for models with tens of billions of parameters. Both hold the same title. The first makes 130k. The second makes 400k. When you plan a study path, you need to be honest about which one you are aiming for, because the training required is very different.

For most career-changers, the realistic on-ramp is applied ML engineering: taking pretrained models, fine-tuning them for a specific business problem, wrapping them in a serving layer, instrumenting monitoring and evals, and shipping. This work pays 140k to 220k at mid-career, which is excellent, and does not require a PhD. It does require solid software engineering (many junior applied ML people fail here, not on the ML), and it requires taste about when ML is the right tool at all. Our AI engineering path writeup goes into the specific curriculum tradeoffs.

Risk factors: the field is currently in an investment bubble, and hiring will contract when interest rates or model-economics assumptions shift. Skills concentrated in a single vendor's platform (say, deep OpenAI ecosystem knowledge with nothing else) carry more risk than transferable fundamentals. Optimize for the fundamentals; the vendor layer will keep churning.

The high-floor field: cloud security

If your money profile is different, high floor rather than high ceiling, cloud security is one of the strongest bets in 2026. The compensation ceiling is lower than ML engineering (top individual contributors cap around 300 to 350k in most markets rather than 500k plus), but the floor is much higher, hiring demand is chronic, and the field is unusually resistant to economic downturns because compliance and breach risk do not care about the macro cycle.

The entry cost is meaningful but predictable. You need Linux and networking fundamentals, cloud provider fluency (AWS or Azure or GCP, pick one and go deep), infrastructure-as-code (Terraform), containers, at least one scripting language, and the security-specific knowledge stack: identity and access management, encryption, threat modeling, incident response, and the major compliance frameworks that shape enterprise buyers' checklists. Certifications matter more in this field than in most others: the AWS Security Specialty, the CCSP, and increasingly the Kubernetes security certs (CKS) actually move salary numbers.

What makes cloud security financially attractive is the shape of the demand curve. Every company running production workloads on cloud needs someone who understands both the cloud and the security implications, and the pool of people who genuinely have both is small. Most "security engineers" hired in the last decade came from pure infosec backgrounds and struggle with modern cloud architectures. Most cloud engineers do not know security in depth. Whoever sits in the middle of that Venn diagram commands a premium, and the diagram is not filling in quickly.

We wrote a detailed piece on how much cloud security engineers make that breaks down the numbers by seniority and region. The short version: at three years of experience, 130 to 170k USD is normal in major North American markets; at seven years, 200 to 260k is normal. In Europe the numbers are 60 to 75 percent of that. In both markets the trajectory is unusually steady.

Downsides: on-call rotations exist, especially at incident-response-focused roles. The work involves a lot of paperwork (audits, control documentation, risk assessments) that engineers who came in expecting pure hacking sometimes find deadening. And the field rewards conservatism, which cuts both ways. It is not the place to be if you want to ship fast and break things.

The stable-ladder fields: data engineering and SRE

Between the high-ceiling frontier and the high-floor specialties, there is a middle band of stable-ladder fields that reward patient competence. Data engineering and site reliability engineering are the two clearest examples. Neither is glamorous. Both pay well and reliably.

Data engineering has quietly become one of the best money-adjusted career choices in tech. The work is unsexy: build pipelines, model warehouses, keep dbt runs green, wrangle Airflow, negotiate with data producers upstream and analysts downstream. The salaries are not unsexy. Mid-career data engineers in the US routinely clear 160k total comp, and the ceiling for principal-level data engineers at scaled companies is comfortably in the 250 to 300k range. In Europe, the same roles pay 65 to 110k euros, with senior roles at scaled companies going higher.

What makes data engineering financially attractive is that the skill stack composes over time. SQL never becomes obsolete. Dimensional modeling instincts learned in 2015 still apply in 2026. Even when the specific tools rotate (Airflow to Dagster, Redshift to Snowflake to Databricks, dbt to whatever comes next), the mental model of how to move and shape data reliably transfers. That means every year of experience is genuinely additive, which is not true in every tech field. If you want a clearer sense of how it differs from adjacent roles, our breakdown of data science vs data analytics vs data engineering walks through the pay and skill divergence.

SRE is the other stable-ladder pick. The work is running production systems at scale, which means capacity planning, incident response, reliability engineering, and automation. The pay is strong (140 to 260k mid-career in the US) and the ceiling is high enough (staff SRE at a scaled company clears 350k) that few people leave for money reasons. The catch is the on-call rotation, which is genuinely taxing and does not go away as you get senior; it just changes shape. If you have young kids or a partner who works nights, be honest about whether you can sustain the rotation for a decade.

Both fields reward the same underlying temperament: comfort with detail, tolerance for boring work done well, a preference for systems that keep running over systems that impress in demos. If that is your temperament, these are excellent money bets. If it is not, you will hate the work no matter what it pays.

The field with the highest hidden variance: full-stack engineering

Full-stack web engineering is the field with the widest gap between what the average salary chart says and what your actual outcome will be. The stated range, 100 to 180k mid-career in the US, is technically accurate. But the variance inside that range is higher than in almost any other field, because "full-stack engineer" is a title that covers everything from junior React developer at an agency to staff engineer at a fintech unicorn.

If you optimize for money inside full-stack, three factors dominate: employer tier, specialization, and market timing. Employer tier is obvious, big tech pays more than a regional consultancy. Specialization is less obvious. A full-stack engineer who is genuinely fluent in a high-value niche (payments, real-time systems, complex data-visualization, distributed systems) makes more than a generalist by 30 to 50 percent. Market timing matters because web engineering hiring is cyclical in a way that data and security are not; when VC funding contracts, full-stack hiring contracts first.

The entry cost is genuinely lower than most high-paying fields. A determined career-changer can be interview-ready for junior full-stack roles in 8 to 12 months, and the bootcamp ecosystem, for all its problems, has a real track record of placing people. But the job market for junior full-stack has been the tightest in tech for two years running, precisely because entry is easy and every bootcamp on earth funnels people into it.

The strategic implication: if you enter full-stack purely for money, you must commit to specialization within two years of your first job. Generalist full-stack engineers plateau early. Specialized ones (payments backend, real-time frontend, ML-adjacent backend, developer tooling) keep compounding. The best of the specialists make money comparable to ML engineers, without the math tax.

Most people who ask us about "which field pays most" and gravitate toward full-stack because it is familiar underestimate this specialization requirement. They imagine that just being a working full-stack engineer will produce the salaries in the chart. It will produce the middle of the chart, not the top. If money is your primary driver and full-stack is your realistic entry, plan the specialization pivot before you take the first job, not after.

Fields to be cautious about in 2026

Some fields still pay well in absolute terms but carry rising risk in 2026, and if money is your primary driver you should weigh that risk carefully.

Pure front-end engineering, especially the "React developer" archetype without deeper systems knowledge, is under compression. AI-assisted code generation eats the routine parts of the work faster than in any other subfield, and the hiring pipeline is oversupplied. The senior roles still pay, but the ladder from junior to senior has fewer rungs than it did five years ago.

Pure data analytics, understood as SQL plus a BI tool plus dashboarding, is another compressed field. It pays fine (80 to 140k) but the ceiling is real and analytics tools with built-in AI assistants are lowering the skill floor for the routine work. Analysts who move up the stack (into analytics engineering, into product analytics with causal-inference chops, into forecasting) do well. Analysts who stay in dashboard-building do not.

QA and manual testing continue their long decline. Test automation engineering (writing frameworks, integrating into CI, defining test strategy) is fine and pays well. Clicking through test scripts is not a career in 2026.

Crypto engineering has become an unusually bimodal field. The people who stayed through the winter and now work on serious infrastructure make excellent money. The broader ecosystem hires and fires with the token price. If you are money-driven, this is a high-variance bet, not a high-expected-value one.

Generic "IT support" as a career is not a money field. It is a legitimate on-ramp to cloud engineering, security, or systems administration, but the money is on the ramp, not on the platform. If you land in support and stay in support, you have chosen a floor of roughly 55 to 85k for the duration.

Finally, be cautious about fields whose demand is heavily tied to one platform or vendor. Salesforce ecosystem work has paid well for years, but the concentration risk is real. Same for Oracle, same for SAP, same for any single-vendor specialty. The money is currently there. Whether it is there in ten years is a bet on that vendor's roadmap and pricing power, not on your skills.

Translating a target salary into a study plan

Here is the mechanical part of the exercise. You have a target salary, either an absolute number ("I need to earn 140k within four years") or a lifestyle-derived one ("I need enough to cover a 3800 dollar monthly mortgage plus expenses, so about 110k gross"). How do you turn that into a specific plan?

First, work backward from the target to the field. Cross-reference the target against the salary bands above, and identify every field where your target is inside the mid-career band (not the top). If your target is 140k, that is inside the mid-career band for essentially every field on the list. If your target is 250k, the list narrows to ML engineering, cloud security, senior data engineering, SRE, and a few others. If your target is 400k, the list is short: senior ML, senior AI research, staff-level SRE at top firms, specific fintech and quant roles.

Second, subtract the fields you cannot afford to enter. Look at the entry cost band for each candidate field and compare to your available runway. If you have twelve months of runway and no prior software experience, ML engineering is not on your list even if the pay would work. This is the step where money-driven planning saves people from expensive mistakes: they see the salary chart, ignore the entry cost, and enroll in a program they cannot afford to finish.

Third, among the fields you can afford to enter, pick based on secondary factors: on-call tolerance, math tolerance, appetite for compliance work, geography flexibility. These matter less than salary in a money-driven plan, but they matter enough to decide between two otherwise similar options.

Fourth, build a concrete twelve-month curriculum for your chosen field, with weekly deliverables and a portfolio project that a hiring manager could reasonably review. Do not enroll in a program without this artifact. If you cannot articulate what you will have built and what you will know by month twelve, you are not making a career plan, you are shopping for hope.

Fifth, plan the interview phase separately from the study phase. Most career-changers underestimate this by six months. Interview preparation is its own skill, and the salary you actually land depends more on interview performance than on GPA or portfolio quality. Our writeup on technical interview preparation covers the parts that generalize across fields, and it is worth reading before you pick a start date for job applications.

The often-overlooked money option: teaching what you already know

One income path that money-first planners consistently ignore is monetizing existing expertise instead of, or alongside, learning a new field. If you already have five years of experience in a specialty (any specialty, not just tech), you are closer to a supplemental income stream through teaching, tutoring, mentoring, or advisory work than you are to a new career.

This matters for money-driven planning in two ways. First, it is a genuine income source. Second, it can bridge the runway problem: if your target field has a 24-month entry cost and you have 12 months of runway, teaching what you know can extend the runway enough to complete the transition. Refonte Learning built its instructor platform specifically to make this bridge accessible, and you can become an instructor on Refonte Learning with an application that focuses on your specific expertise rather than credentials.

The money is real but not fantasy money. Depending on your specialty, hours available, and pricing, teaching or mentoring on a platform typically produces 500 to 4000 USD per month of supplemental income. That does not replace a tech salary. It does buy runway. It also builds the exact skills, communication, breaking down complex ideas, giving feedback, that make you stronger in your target field. Senior engineers who teach are almost always better at code review and design docs than senior engineers who do not.

For career-changers, the sequence often works cleanly: monetize prior expertise through teaching, use that income to fund reskilling into a target tech field, then eventually taper the teaching or convert it into a permanent side stream. For senior tech people already earning well, teaching is more about diversification and long-term optionality than immediate income, but it still compounds financially over years.

The main mistake we see: people who could earn 2000 dollars a month teaching what they already know instead spend those hours doomscrolling and complaining that reskilling is impossible on their budget. The runway problem is often more solvable than it looks.

Risk profile: what "the ladder collapses" means and how to hedge

Salary charts assume the ladder you are climbing continues to exist. Sometimes ladders collapse. The Flash developer market collapsed in 2011. The pure-mobile-app-shop market collapsed around 2015. The blockchain market collapsed in 2022 and partially rebuilt on different foundations. If money is your primary planning criterion, you cannot ignore the probability that your chosen ladder shortens or disappears.

Some fields are highly ladder-stable. Data engineering has been fundamentally the same shape since roughly 2013, and the underlying skills (SQL, distributed systems, data modeling) transfer to whatever the next generation of tools looks like. Cloud infrastructure roles have similar stability. Security roles have exceptional stability because the threat landscape does not go away.

Some fields are ladder-volatile. AI engineering right now is high-paying but the skill mix required has changed every 18 months for the past six years, and the specific tools you master today may be less relevant in three years. That does not mean AI is a bad bet, it means you are paying a hidden tax in continuous re-learning that other fields do not charge. Factor that tax into the salary math.

Some fields are ladder-narrowing. Pure front-end and pure QA are examples. The jobs exist, the pay is not bad, but the ladder does not go as high as it used to, and it is getting shorter each year. If you enter these fields, plan a lateral move within four to six years.

Hedging strategies for money-driven planners:

• Pick fields whose fundamentals are more durable than their tooling (SQL, Linux, networking, distributed systems, security fundamentals, statistics)
• Avoid over-indexing on a single vendor's ecosystem, however lucrative it currently is
• Build enough general software engineering that any specialty can be exited into another
• Keep a teaching or advisory side channel that does not depend on your primary field surviving
• Recognize when you have accumulated enough savings that ladder risk matters less; at that point you can trade income for interest

Refonte Learning's DevOps engineer vs SRE career path piece has more on how adjacent fields hedge each other; the pattern generalizes beyond that specific pair.

Putting it together: three example plans

Abstract advice is easier to reject than concrete examples, so here are three sketches of money-driven plans for different starting points. None are prescriptions; they are calibrated illustrations.

Case A: 26 years old, marketing analyst at a mid-sized company, 62k salary, 18 months of savings runway if lifestyle stays constant, comfortable with spreadsheets and basic SQL, no code background beyond that. Money goal: reach 130k within four years.

Plan: target data engineering. Twelve-month study plan focused on Python, SQL depth, one warehouse (Snowflake or BigQuery), dbt, and Airflow or Dagster, with a portfolio project that ingests a real public dataset, transforms it, and serves an analytics layer. Interview prep months 10 through 14. Realistic first-job offer band: 85 to 110k. Two years in, refactor to a scaled-company role at 130 to 150k. Goal met by year four with margin.

Case B: 38 years old, former teacher transitioning careers, 44k current salary, 6 months of runway, strong communication skills and comfortable with structure but no technical background. Money goal: reach 90k within three years.

Plan: this is the runway-constrained case. Do not attempt ML engineering, data engineering, or SRE; the entry cost exceeds your runway. Target the intersection of two moves: first, monetize existing teaching skills for supplemental income (500 to 1500 monthly) to extend runway to 12 to 14 months. Second, target full-stack web with a specialization intent (backend-leaning, one framework, plus SQL and cloud fundamentals). First-job band 65 to 85k at nine to twelve months. Move to specialized role at 90 to 110k by year three.

Case C: 31 years old, working software engineer at 115k, four years experience in generalist backend, no strong specialization, willing to invest evenings and weekends but cannot take unpaid time. Money goal: reach 200k within three years.

Plan: specialization pivot without leaving current job. Two candidate targets, cloud security or applied ML engineering. Both fit the income constraint (no unpaid runway required) and both compound onto existing engineering skills. Pick based on temperament: if you like careful, methodical work with compliance context, cloud security; if you like modeling, experimentation, and are willing to relearn linear algebra, ML engineering. Twelve to eighteen months of focused evening study on the chosen specialty, then internal transfer or external job move at month 18 to 24, targeting 160 to 190k. Second move at year three to 200k plus.

The common thread across these cases is that money-driven planning is not the same as chasing the highest number on the salary chart. It is matching your target, your runway, and your starting skills to a field whose ladder you can realistically climb, and then executing a specific plan rather than a vague intention.

About Refonte Learning

Refonte Learning is an EdTech platform operated by Refonte Infini Infiniment Grand, a French SAS registered at SIREN 949 841 605 (https://data.inpi.fr/entreprises/949841605), with an operational office at 1 Poulton Close, Dover, Kent, United Kingdom, CT17 0HL. We build training programs in AI, data, cloud, DevOps, cybersecurity, and software engineering, and we work with career-changers, students, and working professionals who need concrete plans rather than motivational content.

If you have specialized expertise in any of the fields discussed above, and you want to convert that expertise into supplemental income while helping the next generation of practitioners, apply to teach on Refonte Learning. Our onboarding is focused on demonstrated expertise rather than credentials, and the platform is built to let practitioners teach the specific narrow slice they know deeply, not force them into generalist survey courses.