Forward Deployed Engineer

Strategy is the act of choosing what to do when you can't know enough to be sure. Manifold builds software for the people making those decisions on behalf of the United States. The ones whose decisions shape the next decade, not the next quarter.

This work is still done largely by hand, by smart people with slide decks, spreadsheets, and instinct. We think it should be done with better tools. Not tools that make the decisions, but tools that sharpen the judgement of the people responsible for them. Cognitive enhancement, in the literal sense.

About the Role

As Manifold's first Forward Deployed Engineer (FDE), you'll spend your week split between writing code and sitting next to the people who will use it. Your job is to close the distance between what operators and action officers actually need and what gets built. The pace is days to a working prototype, weeks to something they rely on.

You are the technical point of contact for these customers, and you have wide latitude on what to prototype for them. Your eyes and ears on the ground will be critical to setting product direction, and you'll have a seat at that table., * Sound technical judgment: You have an instinct for which customer requests are reasonable, which will take real effort, and which aren't worth building. You can size a problem quickly and you're usually right.

• Clear, concise communication: You can explain what you're building, why you're building it, and how it works. To engineers, to action officers, to anyone in the room.
• Bias to action under ambiguity: The spec will rarely be clean. The right next step will rarely be obvious. You move anyway, with enough judgment to course-correct fast when you're wrong.
• Taste in what to build: Anyone can build what a customer asks for. The harder skill is looking at a workflow and seeing the one tool that would actually change it, versus the five tools that would just be moderately useful.
• Curiosity that shows up in the work. You want to understand the mission: how the customer's job actually works, what they're trying to accomplish. You also want to understand the tools (agentic coding, new models, new frameworks) well enough to use them, not just talk about them. The prototype you ship next month should be better than the one you shipped last month because of something you picked up in between.
• Can write code without an LLM. Use the tools, they're good and the job moves fast. But the code you ship is code you understand. When a prototype breaks in front of a customer, or a teammate needs to build on what you wrote, "the model generated it" isn't an answer.

Do you have experience in Software engineering?, * Bachelor's degree in computer science, electrical engineering, math, physics, or a related technical field OR equivalent experience shipping production software

• 2+ years of experience writing production code in a setting where someone other than you had to use, debug, or maintain it.
• U.S. citizenship (for security clearance eligibility), * You've shipped technical work to non-technical customers in a messy environment. You've scoped, built, and delivered something for a customer who couldn't write the spec themselves. You've pushed back on a request when the technical reality didn't match what they wanted. You've done it without a clean problem statement or a fixed set of constraints.
• You've worked with government, defense, or operational customers before. Not required, but it shortens the ramp meaningfully.
• Secret / Top Secret Clearance: Not strictly required, but would help you hit the ground running much faster
• Technical Experience: Prior exposure to Typescript, React Python., * You can stand up a full-stack v0 in 48 to 72 hours: Frontend, backend, deployed, in someone's hands. Not polished, but working.
• You've worked at a small team or early-stage startup: You know what it's like to operate in an ambiguous high ownership environment.
• You build things outside of work: Open source contributions, side projects, things you started because you wanted to see them exist.

4075 Wilson Blvd # 8th, Arlington, VA 22203 Hybrid work $120,000 - $200,000 a year - Full-time, Pulled from the full job description

• Parental leave
• 401(k)
• Health insurance
• Retirement plan
• 401(k) matching
• Vision insurance
• Dental insurance, * 401(k) with matching
• Comprehensive medical, dental, and vision coverage for you and your dependents
• Unlimited PTO
• Health & Wellness stipend
• Company-wide break the last two weeks of the year
• Supportive leave of absence including time off for military service, medical events, and parental leave

Job Type: Full-time

Pay: $120,000.00 - $200,000.00 per year, * 401(k)

• 401(k) matching
• Dental insurance
• Health insurance
• Relocation assistance
• Retirement plan
• Vision insurance

Embed with lighthouse customers to understand their workflows and capability gaps: Plan on weekly contact and biweekly onsite with our customers. The goal isn't just to map what people do. It's to understand why the workflow exists, what it's meant to accomplish, and where Manifold can credibly help. Most of the useful information is stuff people don't think to tell you. Translate operational requirements into technical specifications and constraints: Customers will tell you what they want. Your job is to figure out what they need, what's actually buildable, and what isn't worth building. Pushing back on a feature request is part of the job; so is recognizing the throwaway comment that turns out to be the whole problem. Build and ship workflow prototypes. Own prototypes end to end, from first conversation to deployed tool, including the unglamorous parts: auth, data plumbing, getting it to work on government laptops. The bar isn't a demo, it's something a user opens on Monday and is annoyed to lose on Friday. Equally important: knowing when to throw a prototype away. Most of what you build will not become a product, and recognizing that early is part of the job. Close the loop with the rest of engineering. Bring what you learn back in a form the product team can act on. Not a trip report, but a clear read on what to build next and why. Your signal from the field is one of the main inputs to product direction.