A forward deployed engineer is a software engineer who is sent to work inside a customer's organization, learns how that organization actually runs, and builds or connects whatever it needs, instead of building one generic product from a distance. The role grew up inside enterprise software and AI companies, it is now one of the fastest-growing job titles in technology, and it describes almost exactly the kind of technical help a clinic needs and can rarely buy. This post explains what a forward deployed engineer is, how the role differs from a software engineer, a fractional CTO and an agency, what it costs next to an in-house team, and why the model fits primary care and behavioral health clinics that will never have an engineering department.
What is a forward deployed engineer?
The short definition: a forward deployed engineer, often shortened to FDE, is an engineer who works at the customer rather than at the vendor. "Forward deployed" is borrowed from the military, where it describes units stationed close to where the action is instead of back at headquarters. Applied to software, it means the engineer sits with the people doing the work, inside their systems, and measures success by whether a specific workflow got better rather than by whether a feature shipped.
The role was invented by enterprise data companies in the early 2010s, which had learned that their software only produced value once someone made it work against each customer's messy, specific data, and that this could not be done from an office across the country. The job is often described as close to that of a startup chief technology officer: a small team owning a high-stakes project from start to finish, with the authority to change how the customer works as well as what software they use.
The model has spread quickly. The Financial Times reported that job postings for forward deployed engineers rose more than 800 percent from the start of 2025 through September, as AI companies discovered the same thing: a model or an agent only delivers value once it is wired into each customer's systems and habits. OpenAI, Anthropic, Google Cloud and Databricks all now hire for the title. Google Ads keyword data puts US searches for "forward deployed engineer" above 20,000 a month on average over the past year, and above 40,000 a month in the most recent months, up roughly fivefold year on year.
What a forward deployed engineer actually does day to day
The job is a mix of four things that are usually split across four people. First, the engineer writes production code: connecting an API, fixing a broken sync, building the missing piece between two tools. Second, they scope: they sit with the people doing the work, watch where time goes, and turn a vague complaint ("the booking page is a mess") into a concrete technical problem with a defined end. Third, they integrate inside whatever already exists, including legacy systems, odd data formats, and the compliance rules that come with the industry. Fourth, they carry what they learned back, so the next deployment starts further ahead.
For a clinic, a week of forward deployed engineering looks less like a software project and more like an operations fix. The engineer spends a morning with the front desk watching how a new patient inquiry moves from a directory listing into the EHR, finds that the insurance details are retyped by hand on every booking, builds the connection that carries them across, tests it with real bookings, documents it, and hands it back to the practice.
Forward deployed engineer vs software engineer
A traditional software engineer builds one product for many customers, works from a roadmap, and rarely meets the people using what they build. Their output is features. A forward deployed engineer works for the customer in front of them. Their output is a workflow that works. The skills overlap heavily, which is why most forward deployed engineers started as software engineers, but the orientation is opposite: one optimizes a product for the average customer, the other optimizes a specific customer's operation using whatever products it already has.
That difference is why the distinction matters to a clinic. A clinic does not need a product built. It already has an EHR, a booking page, a clearinghouse, an intake tool, a phone system and a payment processor. What it needs is for those products to work together, and for someone to own the gaps between them. A software engineer hired into a clinic will, by instinct, start building; a forward deployed engineer starts by connecting.
Forward deployed engineer vs fractional CTO vs agency
Clinics that go looking for technical help usually meet two other options. A fractional CTO gives a practice senior technical judgment for a few hours a week, which helps with strategy and vendor choices but usually leaves someone else to do the building. An agency builds, but it typically arrives with a fixed scope, a long contract, and little understanding of how a clinic operates, and it often leaves when the invoice is paid. A forward deployed engineer sits between the two: hands-on enough to ship the integration, close enough to the clinic to understand why it matters, and scoped to the task rather than to a retainer. We compare the first two in detail in fractional CTO vs forward deployed engineer for a medical practice.
| Requirement | Forward deployed engineer | Software engineer | Fractional CTO | Agency |
|---|---|---|---|---|
| Writes and ships production code | Forward deployed engineer: Yes | Software engineer: Yes | Fractional CTO: No | Agency: Yes |
| Works inside your systems, with your staff | Forward deployed engineer: Yes | Software engineer: No | Fractional CTO: Partly | Agency: Partly |
| Learns how your clinic actually runs | Forward deployed engineer: Yes | Software engineer: No | Fractional CTO: Partly | Agency: No |
| Senior judgment on vendors and strategy | Forward deployed engineer: Partly | Software engineer: No | Fractional CTO: Yes | Agency: Partly |
| Paid per task, no retainer | Forward deployed engineer: Yes | Software engineer: No | Fractional CTO: No | Agency: No |
Our reading of the four models as clinics usually meet them. Individual firms and people vary.
What a forward deployed engineer costs, next to an in-house team
Even a small in-house engineering team is expensive. The US Bureau of Labor Statistics puts the median annual wage for software developers at $135,980 as of May 2025. Wages are only part of the cost of an employee: across private industry, benefits make up about 30 percent of total compensation, according to the same agency. Put those together and a single median engineer costs a clinic roughly $194,000 a year, and a minimal team of three, enough to cover integrations, a website, and on-call fixes, comes to about $580,000 before recruiting fees, tooling, or management time.
Those figures assume the clinic can hire well, which is its own problem. A practice owner rarely knows how to interview an engineer, judge their work, or tell whether a project is going badly until it has already failed. The engineer they do hire is usually alone, which means one person carries every system, every password, and every undocumented fix, and the practice is in trouble the week that person leaves.
A forward deployed engineer is priced differently. Because the work is scoped as tasks, a clinic pays for a three-week integration or a two-month migration rather than for a salary that runs whether or not there is engineering to do. The per-hour rate of a senior engineer is higher than the per-hour cost of an employee, but a clinic that needs perhaps eight to twelve weeks of engineering in a year pays for eight to twelve weeks, not fifty-two, and it carries no recruiting cost, no benefits, and no single point of failure.
Where a clinic's money goes, even with venture funding
The reason clinics rarely have engineers is not neglect. It is arithmetic. The easiest way to see it is to look at a well-funded company whose numbers are public. Talkspace is a venture-backed behavioral health company that now trades publicly, so its spending is reported line by line. In its full-year 2025 results, Talkspace reported $228.9 million in revenue. Of that, $130.5 million went to the cost of delivering care, which is mostly what it pays its therapists and psychiatrists, and $53.8 million went to sales and marketing. Research and development, the line that pays for engineering, came to $9.5 million.
| Part | Share (%) |
|---|---|
| Delivering care | 57 |
| Sales and marketing | 23.5 |
| General and administrative | 9.5 |
| Research and development | 4.2 |
| Everything else | 5.8 |
Talkspace full-year 2025 results: revenue $228.9M, cost of revenue $130.5M, sales and marketing $53.8M, general and administrative $21.8M, research and development $9.5M. "Everything else" is the remaining $13.3M.
Talkspace is a technology company at heart, and it still spends more than five times as much on reaching new members as on building its product. A typical clinic tilts much further. Most private practices and community clinics have no research and development line at all. Their costs are clinicians, support staff, rent, billing, and the marketing that brings patients in.
That pattern holds across healthcare practices generally. MGMA, the medical practice management association, reports that labor costs alone take 50 to 60 percent or more of a practice's operating expenses once clinician pay is included, and that practice leaders saw operating expenses rise by about 11 percent in 2025, with staffing costs the most common driver. When the largest cost is already rising that fast, a practice owner choosing between one more clinician and one software engineer will almost always choose the clinician, because the clinician produces revenue the same week they start.
Venture money does not change the choice
It is tempting to assume that a venture-backed clinic group is different. Digital health funding did recover in 2025: Rock Health counted $14.2 billion raised across 482 deals, up 35 percent on the year before. But the money concentrated in a small group of companies, with 26 rounds of $100 million or more. A clinic group that raised a seed or Series A round faces the same arithmetic as a private practice, only with a board watching the growth rate. Every dollar is expected to turn into new patients or new locations, and an engineer is the hardest of those hires to tie to next quarter's numbers.
The technical work clinics keep postponing
Meanwhile the technical work does not go away. It accumulates in the gaps between tools, and it shows up as staff doing by hand what software should do. A few patterns come up again and again in primary care and behavioral health clinics.
- Integrations that stop at the calendar. A directory listing syncs appointments into the EHR, but the insurance details, intake answers, and benefits checks around each booking still get retyped. Our integrations hub shows how many separate systems a typical practice touches.
- An EHR switch that never happens. Practices stay on software that fits badly because moving thousands of records feels too risky to attempt.
- A website that does not convert. The booking page is slow, the intake form breaks on phones, and nobody knows which ads produce booked patients.
- Data that never reaches the clinician. Remote monitoring readings, wearable data, and between-visit check-ins sit in separate apps instead of the chart where they could inform care.
- Security work done late. Access controls, audit logs, and vendor agreements get attention only when a payer, an auditor, or an incident forces it.
None of these is a full-time job on its own. Each is a project with a beginning and an end, which is exactly why a permanent hire is the wrong shape for the problem, and why an engineer who works project by project fits it better.
Why forward deployed engineers fit clinics so well
The forward deployed model matches the way clinics already buy everything else. A practice does not hire a full-time accountant to file its taxes or a full-time attorney to review a lease. It brings in a specialist for a defined piece of work and pays for that work. Technical help has rarely been available on those terms, because most engineering firms are built to sell retainers and long engagements.
Take an illustrative example. Meridian Health Partners is a group with twelve clinicians across two locations, offering primary care alongside behavioral health. Its Series A money is committed to two new clinician hires and a patient acquisition campaign. Over the next six months, the operations lead has three technical problems: new-patient bookings from a directory have to be retyped into the EHR, the practice wants to move to a different EHR before it opens a third location, and remote monitoring readings for patients with hypertension never reach the clinicians who could act on them.
Hiring an engineer would cost Meridian close to one of its planned clinician hires, and that engineer would still need months to learn the practice. Hiring an agency would mean three separate statements of work and a contract that runs past the projects. With a forward deployed engineer, Meridian scopes the three problems as three tasks, starts with the one costing staff the most time, pays for each as it is done, and stops when the work is finished.
How Vybz Health runs forward deployed engineering
The Vybz Health forward deployed engineering service is built around those terms. It exists because we kept meeting clinics that needed real engineering work and had no sensible way to buy it.
- 1
Free technical assessment
An engineer reviews your EHR, website, intake, integrations, and security, and tells you what is broken, what is risky, and what is worth fixing first
- 2
Scope one task
You pick the work that matters most and get a quote for that task alone, with no minimum retainer
- 3
Embedded delivery
The engineer works inside your systems and with your staff until the task is done, usually starting within one to two weeks
- 4
Documented handoff
You get documentation and runbooks, and you own everything that was built, with no lock-in
How a task-by-task engagement runs. The clinic decides what gets built and pays only for the work it approves.
No contract holding the practice in
There is no minimum retainer and no long-term contract. A clinic can engage for a single three-week task, return six months later for the next one, or keep an engineer on a longer project such as a full migration. Because every engagement ends with documentation and a clean handoff, the practice is never dependent on us to understand its own systems.
Engineers who already know healthcare
A general-purpose freelancer spends the first weeks learning what a payer is, why a date of birth matters on a claim, and which data counts as protected health information. Our engineers start with that context, and they have already connected the systems clinics use most, from EHRs to clearinghouses to booking widgets. That shortens every task and avoids the expensive mistakes that come from learning healthcare on a live practice.
Engineers working alongside Luna
Some of the work clinics used to need an engineer for is now handled by Luna, the AI practice manager inside Vybz Health, who works in the EHR, the inbox, and the clinic's other tools every day. A large records move is a good example: our EHR data migration service has Luna move and verify each record, while an engineer handles the parts that need engineering, such as rebuilding a clearinghouse connection or a custom integration. The engineer's time goes to the work only an engineer can do, which keeps each task smaller and cheaper.
Questions to ask before bringing in outside engineering help
Whether a clinic works with us or with someone else, a few questions separate useful technical help from an expensive relationship.
- Can I pay for one task? If the only option is a monthly retainer, the firm is selling capacity, not outcomes.
- What happens when the work ends? Ask to see an example of the documentation and handoff, and confirm that the clinic owns the code, the accounts, and the credentials.
- Have you connected my systems before? Name your EHR, your clearinghouse, and your booking tools, and ask what went wrong last time.
- Who actually does the work? A senior engineer who sells the project and a junior one who builds it is a common pattern.
- How do you handle patient data? Ask about access controls, audit logs, and whether the firm will sign a business associate agreement.
The clinics that get the most from technical help treat it the way they treat their other specialists: clear scope, clear ownership, and the freedom to stop when the work is done. A forward deployed engineer is the first engineering model built to work on those terms, and for practices that will never have an engineering department, it may be the only one that fits.
Also read
- EHR migration: why switching EHR systems is hard, and how Luna handles it: the costs and risks of a records move, and how the work is split between Luna and an engineer.
- Patient intake automation: what Luna does the moment an intake arrives: the intake work that disconnected tools push onto staff, and how much of it no longer needs a person.
- Mental health clinics need an AI employee, not another AI receptionist: why the work between systems is where most of a clinic's operational time goes.
- Healthcare AI agents that get cheaper the longer you use them: the other half of the cost story: how Luna's own memory and learned shortcuts make routine operations cheaper without any engineering.