Palantir co-founder and CEO Alex Karp famously made an observation about why French restaurants are so renowned for their quality: The waiters are an extension of the kitchen staff. They understand how the kitchen operates as well as the cooks, so they can recommend a pairing or custom dish tailored to a diner’s unique taste.
That’s the unlikely culinary inspiration behind the forward deployed engineer, first devised by Palantir nearly two decades ago. The FDE embeds directly with a customer to build the “last mile” of the product to work in production. But unlike a traditional solutions consultant or sales engineer, the FDE is still very much an engineer who writes and debugs production code. Palantir FDEs would spend their days literally onsite (hence the “deployed” in the title) with all kinds of large organizations, from local government agencies to healthcare giants to supply chain operations.
Skeptics wrote off the FDE as glorified consulting for many years, convinced that a true software business shouldn’t need such a labor-intensive deployment process. But Palantir’s outlier success in recent years — it now has a market cap north of $400B — cast new spotlight on the role it pioneered.
If you browse open roles at a startup right now, chances are you’ll come across a listing for a forward deployed engineer. Monthly job listings for the role shot up by 800% from January to September of 2025.
The FDE has taken on a new meaning for AI startups setting their sights on the enterprise (even OpenAI is building out its own fleet). Founders are turning to the FDE model to roll out highly technical AI products to red-tape-lined legacy workflows — an FDE can jump in to build around the blockers that stand in the way of adoption at these types of companies, from unruly codebases to compliance hurdles.
Jake Stauch, co-founder and CEO of Serval (an AI platform for IT) who’s building out an FDE team for his startup, says the model is taking hold at AI startups to help their enterprise customers get value out of agents. “Software platforms have become so powerful that their capabilities are no longer the rate-limiting step for the customer,” he says. “AI unlocked all of these long-tail capabilities, so it can theoretically do anything imaginable. But somebody has to steer the product to do it in that way.”
Given all the demand for FDEs on the startup side, we wanted to dig deeper into what actually makes this role successful. As we found out, while FDEs can be powerful when deployed intentionally, they’re hardly the answer for every company, or customer, and it takes a ton of careful design to get a return on investment in the role.
“Forward deployed engineering is being framed as a panacea right now. But it’s a lot more complicated than that,” says James Honsa, co-founder of the deployment agent company Genera who previously built and scaled Ironclad’s equivalent of an FDE team, called “legal engineering.” “There are times in a company's lifecycle where it makes sense, and there are customer segments where it makes sense, but it's a pretty blunt instrument to try to use for your entire business.”
So we spoke to folks who’ve successfully scaled FDE teams to uncover the exact systems and strategies that worked: the business quirks that justify the steep investment, the hiring strategies that net all-star FDEs, the scope that maximizes the role's impact.
In addition to Honsa and Stauch, here’s our panel:
- Shilpa Balaji, who joined Palantir as an FDE, where she went on to lead FDE recruiting, and is now building an FDE team at Promise, a payment platform for government
- Frank Bien and Lloyd Tabb, former CEO and co-founder of Looker, respectively, who together built and scaled a proto-FDE team
- Tiffany Siu, First Round’s Head of Talent and a former recruiter at Palantir
Let’s dive in.
Where forward deployed engineers add value
The FDE nowadays can get lumped in with an implementation role — but that’s drifted from its Palantir origins, says alum Shilpa Balaji. "Deeply understanding your customer and executing for them through product implementation or configuration is important, but that’s not forward deployed engineering. The FDE model requires making room for creativity and innovation. It’s about discovering new things in a customer context and decentralizing product development,” she says.
Serval co-founder Jake Stauch agrees that FDEs are much more than implementation leads. “The way I see an FDE is as an actual member of the software engineering team. Don’t just force them into implementation. Let them build the software, because they’re the ones talking to customers all day,” he says. “FDEs actually make the product better and more attractive to our ICP, while still reducing friction in the implementation process.”
A Palantir FDE’s charter might be to work alongside a manufacturing customer to reduce the number of defective products coming off of the assembly line, or to deploy software to help a government administer supplies for natural disaster relief. At Serval, for example, which is building an AI ITSM, FDEs have shipped a lot of real product inspired by time spent on the ground with enterprise customers: They’ve built over 60 third-party app integrations, a feedback system for users to rate agent performance and an SLA system within the product.
The most impactful use cases for FDEs typically fall into these buckets, across both product and business.
Push a big deal over the line
Pulling FDEs into the sales process can help close VIP contracts. That’s exactly what Looker did: They’d set up prospects with a free trial, along with heavy pre-sales implementation efforts using their actual data in demos.
“We thought we were deciding between product and service, but not deciding unlocked a third way: Selling as a product and forward-deploying during the free trial so it felt like a customized service,” co-founder Lloyd Tabb told us. “When selling the product, we used the demo as a chance to build a proof of concept, so we didn’t have a dummy sales pitch version — we always asked the prospect for an actual dataset to play with. Then, if we could get our prospects to use the product as much as possible in the free trial, we could comfortably ask for money later.”
Shilpa Balaji shares a similar story from her Palantir days. “We had three or four forward deployed engineers working with a customer in the energy space who just totally hand-rolled something to win the business, to fit the specific problem space, to create something of real value,” she says. “And they weren’t thinking at all about what the roadmap was, or if they had agreement from the mothership. The directive is just build the product, generate value.”
Find obscure but meaty product opportunities by embedding with customers
The “deployed” nature of the role leads to deeper insights that are hard to come by on a 30-minute Zoom customer discovery call. Balaji says FDEs should ultimately be creative problem solvers in the field — who can think without roadmap constraints.
“Living onsite with the customer is such a core part of being an FDE. You're not just setting up a user interview. You're embedding with them. You're prototyping what you hear one day and showing them something the next day,” she says.
That extended immersion is what leads to the most interesting product use cases. “What’s really powerful about an FDE model is that they can be creative. FDEs invent something not by synthesizing it with a broader product, or reprioritizing a roadmap. They’re so scrappily tied to a customer and their problem that they don’t even think about what ‘product’ is,” says Balaji. “Some of the most successful FDE stories from my time at Palantir had nothing to do with the core product offering.”
The value in a forward deployed engineering model is engineers are directly embedded with the customer. And when they see things that other people don't see, they should form conclusions that other people don’t form.
— Shilpa Balaji, former FDE recruitment lead at Palantir
The risk here is that FDEs could crank out a ton of random features that don't improve the core product. So FDEs need a nose for problems that can ultimately serve other customers. “If your forward deploying engineering team has a really strong product sense, chances are that whatever customers ask for, they'll be solving a problem that a lot of other folks have,” says Jake Stauch.
Scale scrappy CTO energy
Another benefit Stauch has found from building out an FDE team is that it recreates the early days of a co-founder single-handedly shipping code for customers.
At Serval, the difference between an FDE and a standard engineer is small: FDEs are expected to spend around 20% of their time with customers, and they’re focused on building product capabilities, and not infra, Stauch says.
“If your forward deployed engineers are the best engineers on your team, you’re reproducing the early co-founder energy where your CTO hears feedback directly from the customer and immediately fixes it and makes the product better. FDEs are a way to scale that,” says Stauch.
The FDE model runs much faster than the traditional feedback-to-product cycle because it removes the “middlemen.” “In any company, you have solutions engineers, customer success staff, account execs. Maybe the solutions engineer hears a good idea and they communicate it to a product manager. The product manager talks to the engineering manager, and then it gets scheduled in a sprint in the quarterly plan,” he says. “But there’s this gap between a customer saying, ‘I wish it worked this way,’ to the problem being addressed. That can take months, even at a great company. So it’s much better if an engineer hears that feedback directly, then goes back to their desk and builds it.”
FDEs recreate what happens in the early days of a startup when it's just a couple founders asking customers, ‘What do you want? Cool, we'll build it.’ And then they’ll come back the next day and say, ‘Did this fix your problem? What else do you want?’”
— Jake Stauch, co-founder and CEO of Serval
Clean out the feature garage
The tighter feedback loop that leads to increased shipping velocity allows FDEs to build lower-priority but still impactful features that might otherwise get stuck in roadmap purgatory.
Before founding Serval, Stauch was Product Lead at physical security platform Verkada. The sales team had a Slack channel called “Feature Garage,” where they’d dump customer feature requests that, largely, went untouched. But Stauch says that with an FDE model, engineers can action on these requests — not by building the exact features, but by addressing the deeper need.
“Forward deployed has a huge role in driving progress on non-controversial product capabilities. No one needs to have a brainstorming session or a roadmap around when to slot that in. They can just go and build it,” he says.
These seemingly small product blemishes can add up. “Overprioritizing is actually a mistake that really good product leaders make,” says Stauch. “They never touch the P2s. But P2s stack up, and if you never have anyone look at P2s, you’ll end up with an inferior product, even though you were technically focused on all the right things. With the forward deployed model, P2s actually get looked at.”
Take this diagnostic before adding forward deployed headcount
While these benefits would entice any founder, FDE dreams can be misguided if a few key business ingredients are missing. “I think when folks today are looking to build out FDE teams, a lot of what they're looking for is just stronger customer signal and faster iteration. But you don’t need a whole fleet of FDEs to do those things,” points out Shilpa Balaji.
Investing in an FDE team, especially in the earliest stages of a startup, is a costly bet — and can quickly burn a lot of cash if the math doesn’t add up.
Frank Bien was tasked with scaling Looker’s scrappy sales motion when he stepped in as CEO. “When I joined, there were still a lot of unknowns. Should we sell for $500 a month or $5,000?” Bien told us. “It was all up in the air — and there were no spreadsheets or slide decks to be found. In my mind, it was similar to Marketo’s model, meaning it wasn’t going to be just a couple hundred bucks a month. We needed to be in the neighborhood of $25,000 a year per customer.”
Validating there’d be a healthy return made Bien confident to invest in an FDE model in the early days. “After crunching the numbers, we saw that by the time we had 2,000 customers, we could be doing $100 million dollars in ARR, and on the path to going public. That was the model from 2013 on,” says Bien.
“You have to know your model backwards and forwards to make a bet like that. We knew the margins on the costs we were sinking into pre-sales support made sense,” he says. “But if we had been unsure whether it was going to take 2,000 customers or 100,000 customers to reach the $100 million run rate, that would have been an incredibly risky move — we very easily could have been lighting our VC dollars on fire.”
If you’re confident in your model, you don’t have to worry whether you can support all the resources you’re sinking — you can do the math and find out.
— Frank Bien, former CEO of Looker
So before putting out a call to FDEs, make sure your business has these three features.
1: You’ve landed (or are going after) big fish
“Forward deployed engineering is definitionally an upmarket motion,” says James Honsa (former head of legal engineering at Ironclad). “You should not be doing this if you believe the end shape of your product is some sort of product-led growth freemium fit.”
That doesn’t necessarily mean you can’t also have customers with smaller annual contract values (ACVs). But the upper end of your customer base should include the likes of the Fortune 500, and that's where a forward deployed team should be spending their energy.
Honsa joined Ironclad before it reached $1M ARR as the first dedicated post-sales hire. He says the early Ironclad team understood that their best business would be with the legal departments of huge global companies — and they’d need dedicated engineering talent to win those contracts. “The very first version of Ironclad’s product worked out of the box. But the founders knew that all the juicy, high-value legal work would require very bespoke implementations. So to lock in these enterprise ACVs, we had to accept that the product couldn’t have pre-made building blocks,” he says.
2: You aren’t prescriptive about how you want customers to use your product
Shilpa Balaji advises founders to ask this question to evaluate their startup’s need for an FDE: How opinionated are you about what your product should be in the future?
If you are very opinionated about what your product should be, FDE might not be a good fit, because so much of their value comes from discovering new product opportunities. “If you do have a strong opinion, or want to, then what you're probably saying when you want an FDE is that you just want more customer signal, or more customer proximity to validate your hypotheses. That’s useful, but a PM could do that, or an engineer who can talk to a customer can do that,” she says.
Balaji explains that, on a product opinionated-ness spectrum, you have Apple at one end and Palantir at the other. Apple’s products are the epitome of “out-of-the-box” — everyone more or less experiences and uses an iPhone the same way. SaaS businesses tend to fall into this camp, too. Palantir, on the other hand, develops platforms that can shapeshift to meet the needs of highly variable problems and organizations.
Balaji says that Palantir’s roadmap in the early days was shockingly uncharted. “Early on at Palantir, we rarely said, ‘This is what the product should be.’ FDEs helped us build incrementally more valuable products. Each thing we built, we had learned from a concrete use case and customer,” she says. “So for a more SaaS-ified product offering, where the founder has a strong and unwavering vision for what that product should be in the future, that tells me they’re not the best fit for an FDE.”
I think the learning the industry is taking away from the FDE’s newfound popularity is, ‘Oh, you should listen to your customers!’ Of course you should, but FDEs are for organic product growth.
— Shilpa Balaji, former FDE recruitment lead at Palantir
3: You don’t have a uniform ICP
We’ve catalogued here on The Review how many startups have found product-market fit by honing in on an ultra-specific ICP. But having a detailed list of customer criteria might actually make you a bad candidate for an FDE team, at least early on.
“FDEs make a ton of sense when your product can be deployed across a diverse range of use cases. Palantir is the extreme, canonical version of this, but at Ironclad, our version of this problem was that we could sell to any industry in the world from our earliest days,” says James Honsa.
Ironclad’s early customer mix had little in common. “For our first 50 customers, we had public tech companies, YC startups, global beauty brands and professional sports teams. We were implementing contract processes for all of them, but the needs for building out an influencer agreement workflow in Japanese were super different from season ticket sales contracts for an MLB team,” he says.
But the one through line of customer personas that warrant an FDE, says Shilpa Balaji, is that they all have technically demanding problems that can’t be solved with out-of-the-box products. “FDEs aren’t just sales engineers. They should be tackling really difficult problems that are very customer-specific, problems that need to be solved to achieve the outcome both you and your customer want,” she says. “At Promise, we’re building out an FDE team because our customer base — the US government — is heterogeneous. Each state administers their government programs differently, and the customer and technical landscapes vary a lot. So we need to be building and learning at the edges.”
Don’t shove an FDE into an engineering or post-sales role
First Round’s Head of Talent, Tiffany Siu, has kicked off searches with founders who start off by saying they want an FDE — only to realize, upon digging deeper, what they actually need is either a formal software engineer or a post-sales role, like an implementations consultant or customer success manager.
“When founders hire an FDE, they often imagine an all-in-one person who can build the product, implement it, train customers, customize it and keep everyone happy. That’s not realistic at scale, but it does capture the need for someone who can bridge engineering and customers early on,” she says.
Siu likes to ask these questions to help founders clarify whether an FDE is the right role to hire for:
- What triggered this opening? Who’s doing this work on your team today? What would happen if you don’t hire this person? “I find these questions help founders frame the real-world use case of this person. Because on paper, sometimes they'll say, ‘We need this person to do this,’ but it doesn't line up with what actually triggered the hire,” says Siu.
- What would this person’s day-to-day work look like? Thinking about how you’d structure a potential FDE’s schedule can help sort out whether this role is truly a customer-facing engineer, or just extra hands for GTM or operations. “It can be surprisingly hard for founders to think of a hire’s schedule on a granular level. Sometimes they’ll say something vague like, ‘I want them to work with customers,’ but getting more specific about what their week would look like can be very revealing,” she says.
- How will you measure this person’s success? “If this person were to crush it or not crush it, what would they have done? Thinking in numbers and key metrics can help you get really clear about the role you need, like quota, user adoption and usage, for example,” she says.
She shares an example of a founder who initially thought they wanted an FDE, but eventually landed on an implementations lead: someone focused on getting the product set up and live for customers. “As we walked through what this person would be doing day to day, it became clear they didn’t want someone to build or extend the product by writing code. They needed someone who could execute repeatable deployments and work closely with sales — someone who could reliably get customers live using the existing product,” she says.
If a founder says, ‘I want this forward deployed engineer to have closed X deals or run X number of demos,’ they probably want someone closer to sales, not an FDE.
— Tiffany Siu, First Round’s Head of Talent
Hiring the right forward deployed engineer
So you’ve sorted out that an FDE does make sense for your startup. How do you find the best person for the job?
You don’t necessarily need to start by scouting someone who’s already been an FDE. “There are many different versions of forward deployed engineers today. An FDE at Palantir can look very different from an FDE at an early-stage startup. So founders shouldn’t optimize for the title alone — it’s worth taking the time to understand how the role is actually structured at that company, and asking candidates what their day-to-day work really looks like,” says Siu.
In some cases, titles and pedigree can even be an anti-signal. The most stellar FDEs our panel has worked with all had these five qualities in common.
Look for these five traits
They don’t bring in a playbook (and are often early in their career). At Palantir, recent college grads made up a bulk of the FDE roster. “An FDE isn’t somebody who brings a playbook with them. They’re not doing a lot of pattern matching. They’re outcome-oriented, independent thinkers, who have a belief that any problem they confront can be figured out,” she says. “New grads bring a fresh pair of eyes to the table. People earlier in their careers just tend to be more open-minded about the problems they solve, and how to solve them.”
A lower seniority level wasn’t a fixed requirement, but generally speaking, the more specialized experience an FDE candidate had, the less likely they’d have that same level of independent thinking. “If someone came in overly dogmatic or set in their ways, I found that was actually a red flag for an FDE. So folks who spent more than 10 years at a FAANG company, for example, were in the no fly zone,” says Balaji.
They’re gritty. Balaji says grit is probably the hallmark quality of the best FDEs she worked with at Palantir. To put it bluntly, she says they had a “willingness to eat pain.” “Forward deployed engineering is painful. So many of the problem spaces FDEs worked in are extremely difficult, so these folks really need to believe they can do the impossible,” she says.
James Honsa agrees, saying that the best FDEs at Ironclad were “grinders.” “There's just a lot of work to do at a hypergrowth company. In deployments, a lot of this work was necessary, but not strategic, and that was a persistent challenge to manage as our customer base grew,” he says.
That run-through-walls energy is a trait shared by founders. Jake Stauch says FDEs often serve as mini-founders in the org. “FDEs have a strong product sense and an actual interest in sales. They have to like the fact that they’re working to get a deal done. These folks are often former or future founders,” he says.
They clear a high technical bar. “Engineer” is still an essential part of the job title. Honsa emphasizes that many of the folks who made the best FDEs at Ironclad could have also gone on to become a staff engineer at a top tech company. “The role is technical. In the early days at Ironclad, legal engineers were merging code into production and doing code review with our CTO every single week,” he says.
Siu agrees that what Palantir got right with their FDE model was never compromising on technical skills. “I think why Palantir was so successful with their FDEs in the early days is that they had to pass the same interview loops and facets as software engineers. They could be a traditional engineer if they wanted to, but they had these other skills that made FDE compelling,” she says.
They don’t stop shipping. “The best FDEs I worked with were compulsive builders,” says Balaji. “This type of person just can’t help but create something. They’re prolific, whether it’s building a tool, releasing some app, or contributing to an open-source project, they just have a restlessness to build.”
They’re deeply curious about how businesses work. “FDEs are innately curious about businesses,” says Honsa. “They’re someone who gets energy from going super deep on the legal risk of influencer marketing and how to create a high throughput process that protects our customers, for example. We firmly believed that we could upskill people who had the right drive.”
Would you want to be in the trenches with this person? That’s the bar I hold for a really exceptional forward deployed engineer.
— Shilpa Balaji, former FDE recruitment lead at Palantir
Interview for raw problem-solving ability (but don’t skip the coding test entirely)
Shilpa Balaji says Palantir diverged from the classic big tech coding interview. Instead, Palantir oriented FDE interviews around the real technical challenges the company was trying to solve for their customers.
“There’s the Google style, which is interview, coding test, interview, coding test, repeat. But lots of our questions at Palantir were oriented around really high-level problem solving,” she says. “So the FDE hiring manager would say, ‘Here’s something that one of our customers is working on. No one has ever been able to solve this problem. How would you solve it?’”
Balaji offers an example of a problem-solving scenario Palantir would share in interviews: “We would explain insider trading to the candidate and ask them to design a solution: What data would they need, what questions would they ask the customer, what would they look for? We wanted to assess both their business reasoning and technical reasoning,” she says.
Honsa says Ironclad similarly set up interviews to screen for problem-solving skills. “Our most effective interview tactic was to ask folks to present a problem from their career and teach us how they used technology to solve it,” he says. “This was intentionally open-ended but elicited amazing responses. One of our founding legal engineers had an amazing presentation where he showed photos of a physical ‘deal room’ from a multi-million dollar aviation financing transaction, and how he’d printed thousands of pages and individually sticky-noted them, followed by the Excel macros he created to automate future closings.”
We looked for forward deployed engineers who got energy from a very open-ended prompt. Because often our customers came to us with very open-ended prompts.
— James Honsa, former Head of Legal Engineering at Ironclad
Scoping the role
Once you’ve embarked on hiring your first FDEs, lean on these three tactics to set up the role for impact — for both the FDEs themselves and your business.
Reserve FDEs for your biggest customers’ hardest problems.
Ironclad didn’t send legal engineers out to onboard every single customer. James Honsa says the team learned over time to reserve their FDEs for only VIPs. “In the early days, we justified the model for every customer. But as we matured, we actually bifurcated our post-sales team into legal engineers, who we reserved for high ACV customers,” says Honsa. “I think teams today make the mistake of thinking you’re either fully an FDE company, or you’re not. Ironclad was successful at having a wide range of menu options given the customer need, and systematizing our downmarket implementations to be more cookie-cutter.”
One of our biggest learnings at Ironclad was figuring out how to sprinkle FDE on the right customers at the right times as we grew.
— James Honsa, former Head of Legal Engineering at Ironclad
Jake Stauch says the FDE’s scope has evolved similarly at Serval. “At first they deployed to all customers. Now we prioritize our largest customers. For us, those are companies with more than 1,000 employees, where it’s more likely they’ll need to build custom capabilities,” he says.
FDEs do their best work when they’re onsite.
Back when Shilpa Balaji was an FDE herself, she says some of her most impactful work happened during extended travel to a customer’s site. “I once spent weeks in a small German town with just two other colleagues. We’d go onsite to the customer’s factory floor. So we had to embrace not just the list of requirements that this customer put in their contract, but we had to spend time with their day-to-day experience. What you discover onsite is going to be so different from what was sold in the contract,” she says.
James Honsa says Ironclad’s FDEs also embraced the travel lifestyle. “One of my favorite moments was when a Fortune 100 General Counsel lovingly called us ‘The Backpacks’ when we arrived at their office,” he jokes.
Embrace scope creep, but be wary of throwing human labor at a product problem.
Leaning into the services nature of the FDE role is a key part of what makes it so impactful — so long as you’re not letting FDEs sink bottomless hours into a solution that won’t benefit future customers.
“In the early days at Ironclad, our view was that if we were experiencing scope creep, it was because the customer had more problems that we could solve,” says Honsa. “The key thing we got right was that our software economics benefited from scope creep. So we priced based on workflows early on. For our second customer, we’d meet a new attorney who said, ‘Hey, by the way, I do this master services agreement workflow. Can Ironclad help with that too?’ We’d say yes, and our sales team would figure out how to monetize it.”
The bad version of scope creep, says Honsa, is conducting the n-th round of iteration on a workflow with capped user volume. “If we found ourselves working on use cases where we were just endlessly iterating, and there was no clear path toward a software upside, we’d have to find a way to break off the work,” he says.
But Honsa says a truly gritty FDE will embrace scope creep in pursuit of the solution to a tough problem. Wrangling the scope should be their manager’s problem — stepping over contractual bounds probably won’t even occur to a great FDE.
“We had a team that was so driven by solving customer problems that they wouldn't even realize it. Sometimes I'd tell my team, ‘Oh, my God, you spent how many hours optimizing this thing?’ and I’d have to get on a call with the customer and rebalance the scope. But that was a good problem to have.”