All roadmaps

Forward Deployed Engineer

Full-stack + AI + customer skills - shipped inside the client's walls.

The FDE path: an experienced-developer role that embeds with customers to deliver working AI systems. Engineering pillars, applied AI, cloud delivery, and the discovery, business and communication skills that make the field half of the job.

Your progress0/53 topics · 0%

Progress is saved in this browser.

8 stages53 topics₹0 to follow
Skill certification coming soon

Filled nodes are the focus for this track; outlined ones can wait. Tap a topic to open it and mark it done, learning or skipped.

Forward Deployed Engineer in 2026Related roadmapsAI Engineer roadmapFull Stack roadmapAI Agents roadmapProve it as you goReading is half of it. The free, scoredAI Engineer certification covers theground this roadmap leads to, and the AIEngineer job simulation puts it to workon real tickets - evidence a recruitercan check. Mark topics done, learning orskipped as you go; your progress carriesinto the classic view.AI Engineer certificationJob simulationClassic viewKnow the job before the toolsThe Forward Deployed Engineer jobin 2026Which skills to learnFull-stack plusAICustomer-facingengAgents in the fieldWhich role to prepare forFDE vs SE vsconsultantFresher vsswitcherWho hires FDEsWhat companies expectThe hiring barWhat interviewstestProof they verifyWhat gets you hiredA 90-day planProjects thatget readCertification and proofUnderstand a role that didn'texist five years ago.IntroductionWhat an FDE isFrom X to FDERoles & responsibilitiesWhere FDEs workFull-stack fluency is the entryticket.Engineering FoundationFull-stack developmentLinux & shellDSA - the interview gateSystem designThe systems customers areactually buying.AI Engineering SkillsLLM fundamentals & modelchoiceRAG & vector databasesAgent architecturesEval pipelines & regressiontestingLatency, cost & inferenceoptimisationAI coding toolsDeploy into environments youdon't control.Cloud & Delivery OpsOne major cloud, fluentlyContainers & infrastructureas codeSecurity, privacy &complianceObservability in the fieldTurn a vague executive wishinto a buildable plan.Discovery & ScopingRequirements gatheringTechnical scoping &sequencingTrade-offs: scope, speed,qualityRapid prototypingSpeak the customer's languageof value.Business AcumenEnterprise workflowsROI & AI impactStakeholder managementProduct feedback loopThe multiplier on everythingabove.CommunicationTechnical writingDemos & presentationsExpectation managementFresher or experienced?Both tracks hold every topic. On the Fresher track thefoundations are filled and the later stages are outlined - dothem when a project needs them. On the Experienced track thefoundations are outlined and the later stages are filled. The2026 section at the top applies to both.Prefer a list?The classic view shows the same 7 stages as a sequentialchecklist with time estimates. Your marks are shared betweenthe two views.Open the classic view

Everything on this roadmap

Every stage and topic on the map, in reading order. Click any topic to open it.

2.Understand a role that didn't exist five years ago.

The Forward Deployed Engineer roadmap, topic by topic

Every topic on the map in reading order, with what to learn and the resources we recommend. Main topics are open; open any subtopic to read it. Select a topic name to open it in the topic panel, where you can track your progress.

  1. 1. Know the job before the tools

    The Forward Deployed Engineer job in 2026

    2026

    The forward deployed engineer is the fastest-growing engineering title of the AI era: a software engineer who sits with the customer, understands their problem and builds the working solution on top of an AI product, part engineer, part consultant, part product manager. Analyses of job postings in 2026 found forward deployed engineer listings up more than 1,000 percent year on year and more than 4,600 percent against 2023, and in September 2026 OpenAI, Anthropic and Palantir were each listing many times more forward-deployed roles than sales roles. LinkedIn's Jobs on the Rise 2026 ranked AI engineer the fastest-growing US title with postings up 143 percent. The four nodes answer which skills, which role, what companies expect and what gets you hired.

    Which skills to learn

    A forward deployed engineer needs full-stack engineering plus applied AI, the customer-facing skills of a consultant, and the ability to put agents to work inside a customer's systems safely. The sub-nodes order the three.

    Full-stack plus AI

    The engineering half is a full stack engineer's: a typed front end, an API, a database, integrations with the customer's systems, deployment and monitoring - built fast, because the customer is waiting. The AI half is the AI Engineer roadmap: model APIs, prompt and context engineering, retrieval over the customer's documents with permissions, structured outputs, evaluation sets that prove the solution works on their data, and cost control. The Stack Overflow 2025 survey's stack - TypeScript, React, Node, Python, Postgres - is the one most forward deployed teams use. Learn both halves to deployment; the role is the combination.

    Customer-facing eng

    The customer-facing half is what makes the role rare: discovering the real problem through interviews rather than the stated one, scoping a solution the customer will adopt, demonstrating progress weekly, writing clearly for executives and operators, handling objections and the moment the model is wrong in front of the client, and leaving documentation the customer can run. Posting analyses in 2026 describe the role as engineering plus product plus consulting, and the Work Trend Index's finding that 71 percent of leaders prefer AI-skilled candidates is strongest here, where the candidate must also explain the AI to the customer.

    Agents in the field

    AI

    Forward deployed work in 2026 is increasingly agents inside customer systems: workflows that read a customer's data, call their tools and take bounded actions with human approval. Gartner expects a third of enterprise applications to include agentic AI by 2028 and warns that more than 40 percent of agentic projects will be cancelled by 2027 for cost and unclear value - the forward deployed engineer is the person who prevents that by scoping the agent to a job that pays off and bounding it with checkpoints, evaluation and logs. Microsoft's 2025 Work Trend Index found 46 percent of leaders already using agents to automate workflows; someone has to deploy them well.

    Which role to prepare for

    The title overlaps with solutions engineer, implementation engineer and consultant but differs in what you build and own. The sub-nodes compare them, lay out the fresher and switcher routes, and say who hires forward deployed engineers.

    FDE vs SE vs consultant

    A forward deployed engineer builds the customer's solution on the product and owns its success; a solutions or sales engineer demonstrates and scopes but usually does not build the production system; an implementation engineer configures rather than engineers; a consultant advises and may build but does not own the product's roadmap feedback. The forward deployed role is closest to a founding engineer embedded with one customer at a time, which is why the companies hiring it - the AI labs, Palantir, Databricks, Scale and a growing set of AI startups - pay it like senior engineering. LinkedIn's Jobs on the Rise 2026 ranked AI engineer the fastest-growing US title with postings up 143 percent; the forward deployed role rides the same demand.

    Fresher vs switcher

    The role is rarely a fresher's first job: most postings want shipped production software and comfort in front of customers, so the fresher route is a full stack or AI engineering role with customer contact for a year or two, or a startup where you did both. Switchers from consulting, solutions engineering or product management with real engineering skill are the most natural fit, and the surge in postings - up more than 1,000 percent year on year in 2026 analyses - has made the switch common. HackerRank's 2025 report found junior hiring the hardest hit, which is one more reason the first job is usually not this one. On this map freshers get the foundations filled and the advanced stages outlined; experienced learners get the reverse.

    Who hires FDEs

    The AI labs - OpenAI, Anthropic - Palantir, Databricks, Scale AI, Google and Stripe appear in the 2026 posting analyses as the largest hirers, each listing far more forward deployed roles than sales roles, and AI startups selling to enterprises follow them; India's global capability centres and AI services firms have begun posting the title for customer deployments too. The work is often on site or in the customer's time zone, which favours candidates near enterprise customers or willing to travel, and it pays well above standard engineering for the same years of experience because so few people have both halves.

    What companies expect

    Companies expect an engineer who can walk into a customer's problem, understand it in days, build a working solution on the product with real data, and leave the customer better off - with the judgement to say what AI should not do. The sub-nodes cover the bar, the interview and what gets verified.

    The hiring bar

    The bar is shipped software and customer judgement at once: a working solution on real customer data within weeks, built on the product with the right AI pattern, evaluated so the customer can trust it, documented so they can run it, and scoped so it pays for itself. The Work Trend Index's 66 percent of leaders who will not hire without AI skills is the floor here; the role also wants the person who tells the customer what AI should not do.

    What interviews test

    Expect a practical build on a realistic customer problem with model APIs, a system design conversation about integrating with legacy systems, a scoping exercise where you interview the interviewer as the customer, a presentation of your solution to a non-technical panel, and questions about a project that failed and what you learned. HackerRank's 2025 finding that practical assessments are the standard is doubly true here: the interview is a simulated engagement.

    Proof they verify

    Hiring teams verify shipped work and customer contact: a deployed product with real users, a repository that shows AI patterns built properly - retrieval, structured outputs, evaluation, cost control - a write-up of a customer or stakeholder project with the outcome, and clear written communication. OneRoadmap's AI Engineer certification and job simulation are free, verifiable evidence of the AI half; the Full Stack certification covers the engineering half.

    What gets you hired

    Shipped software with real users, evidence of building on model APIs, a track record of working with customers or stakeholders, and clear writing. The sub-nodes give the ninety-day plan, the projects that get read and the proof that counts.

    A 90-day plan

    Month one: a full stack application deployed with real users, even a handful, and the AI Engineer roadmap's model API, prompting and structured output stages. Month two: a retrieval-augmented feature over a real document set with an evaluation set and a cost note, built for a real customer - a small business, a nonprofit, a team at your current employer. Month three: a bounded agent with human approval added to it, a written case study with the customer's outcome, the AI Engineer certification, and twenty targeted applications a week to the companies named above.

    Projects that get read

    Projects that get read are deployments for someone else: a document-answering tool for a real organisation with an evaluation set and the adoption numbers; an agent that automates one workflow for a small business with approval steps and logs; an integration between a customer's legacy system and a model-backed service; a case study that states the problem, the solution, the measurement and what the customer did next. Gartner expects a third of enterprise applications to include agentic AI by 2028; the forward deployed engineer is who makes that work for one customer at a time. Posting analyses in 2026 describe the role as engineering plus consulting; a project with a real customer shows both.

    Certification and proof

    Proof is a customer who was helped and the software that did it: a case study, a deployed system, a repository with the AI patterns done properly, a verified certification and clear writing. With forward deployed postings up more than 1,000 percent year on year in 2026 analyses and candidates with both halves scarce, the ones who get replies are the ones who can show a real deployment.

  2. 2. Understand a role that didn't exist five years ago.

    Introduction

    Understand a role that didn't exist five years ago.

    Time: about 2-3 days. Topics: What an FDE is · From X to FDE · Roles & responsibilities · Where FDEs work.

    What an FDE is

    Optional on the Experienced track

    A forward deployed engineer builds and ships production systems inside a customer's environment - half engineer, half consultant, fully accountable for the outcome. Popularised by Palantir, now core to how AI labs and enterprise-AI companies deliver.

    From X to FDE

    Optional on the Experienced track

    This is not a beginner role: most FDEs step in from full-stack, backend or ML engineering. The gap to close is rarely code - it's delivery under ambiguity and customer trust.

    Roles & responsibilities

    Optional on the Experienced track

    Discovery, scoping, building, deploying, demoing and iterating with the customer in the room - plus feeding what you learn back into the core product.

    Where FDEs work

    Optional on the Experienced track

    AI labs, enterprise-AI vendors, data platforms and consultancies - teams selling outcomes, not seats, hire FDEs.

  3. 3. Full-stack fluency is the entry ticket.

    Engineering Foundation

    Full-stack fluency is the entry ticket.

    Time: about ongoing. Topics: Full-stack development · Linux & shell · DSA - the interview gate · System design.

    Full-stack development

    Frontend, APIs, databases and auth - FDEs deliver complete apps alone, often in weeks. If any layer is a black box, close it first.

    Linux & shell

    Customer environments are Linux boxes, VPNs and locked-down shells - comfort here is non-negotiable in the field.

    DSA - the interview gate

    FDE interviews still start with algorithms; you won't get to show AI skills if you don't pass this round.

    System design

    You must prove to customers that what you built won't break or go rogue in production - design reasoning is the proof.

  4. 4. The systems customers are actually buying.

    AI Engineering Skills

    The systems customers are actually buying.

    Time: about 4-6 weeks. Topics: LLM fundamentals & model choice · RAG & vector databases · Agent architectures · Eval pipelines & regression testing · Latency, cost & inference optimisation · AI coding tools.

    LLM fundamentals & model choice

    Capabilities, context, cost and data-privacy posture across providers - choosing a model per customer constraint, not preference.

    RAG & vector databases

    Grounding models in the customer's documents with retrieval - the default architecture of enterprise AI delivery.

    Agent architectures

    Tool use, memory and state, multi-agent orchestration and MCP - matched to the workflow being automated, with the simplest design that works.

    Eval pipelines & regression testing

    Golden sets built from customer data, regression runs before every demo - evals are how you prove the system to a skeptical stakeholder.

    Latency, cost & inference optimisation

    Caching, routing and token budgets - enterprise deals die on unit economics as often as on accuracy.

    AI coding tools

    Claude Code, Cursor and agentic workflows - FDEs use AI leverage to hit field timelines nobody else can.

  5. 5. Deploy into environments you don't control.

    Cloud & Delivery Ops

    Deploy into environments you don't control.

    Time: about 3-4 weeks. Topics: One major cloud, fluently · Containers & infrastructure as code · Security, privacy & compliance · Observability in the field.

    One major cloud, fluently

    Compute, networking, IAM and managed databases on AWS/GCP/Azure - customer deployments live inside their cloud accounts and their rules.

    Containers & infrastructure as code

    Docker and Terraform so every customer environment is reproducible - snowflake deployments are field debt.

    Security, privacy & compliance

    VPC isolation, data residency, API security and audit trails - passing the customer's security review is part of shipping.

    Observability in the field

    Logging, tracing and dashboards the customer can read - the system must be debuggable after you fly home.

  6. 6. Turn a vague executive wish into a buildable plan.

    Discovery & Scoping

    Turn a vague executive wish into a buildable plan.

    Time: about 2 weeks. Topics: Requirements gathering · Technical scoping & sequencing · Trade-offs: scope, speed, quality · Rapid prototyping.

    Requirements gathering

    Interviewing users, shadowing workflows and finding the real problem behind the stated one - the highest-leverage hours of any engagement.

    Technical scoping & sequencing

    Slicing delivery into demos that build trust: first value in days, not quarters.

    Trade-offs: scope, speed, quality

    Saying 'we can have two of three, here's my recommendation' - out loud, early, in writing.

    Rapid prototyping

    Clickable proof in the first week - prototypes settle debates that meetings can't.

  7. 7. Speak the customer's language of value.

    Business Acumen

    Speak the customer's language of value.

    Time: about 2 weeks. Topics: Enterprise workflows · ROI & AI impact · Stakeholder management · Product feedback loop.

    Enterprise workflows

    Procurement, security reviews, change management and who actually owns budget - the invisible machinery your project moves through.

    ROI & AI impact

    Quantifying hours saved, error rates cut and revenue unlocked - the numbers that renew contracts.

    Stakeholder management

    Sponsors, users, IT and skeptics each need different messages - mapping and serving them deliberately.

    Product feedback loop

    Field learnings routed back to the product team - FDEs are the company's highest-bandwidth sensor.

  8. 8. The multiplier on everything above.

    Communication

    The multiplier on everything above.

    Time: about ongoing. Topics: Technical writing · Demos & presentations · Expectation management.

    Technical writing

    Optional on the Fresher track

    Scoping docs, decision records and runbooks the customer keeps after you leave - writing is the deliverable's other half.

    Demos & presentations

    Optional on the Fresher track

    Narrated, rehearsed, failure-proofed demos - the heartbeat of field engagements.

    Expectation management

    Optional on the Fresher track

    Delivering bad news early with options attached - trust compounds from honesty under pressure.

Prove it as you go

Reading is half of it. The free, scored AI Engineer certification covers the ground this roadmap leads to, and the AI Engineer job simulation puts it to work on real tickets - evidence a recruiter can check. Mark topics done, learning or skipped as you go; your progress carries into the classic view.

Fresher or experienced?

Both tracks hold every topic. On the Fresher track the foundations are filled and the later stages are outlined - do them when a project needs them. On the Experienced track the foundations are outlined and the later stages are filled. The 2026 section at the top applies to both.

Prefer a list?

The classic view shows the same 7 stages as a sequential checklist with time estimates. Your marks are shared between the two views.

Frequently asked questions

What is a forward deployed engineer?

A software engineer who sits with the customer, understands their problem and builds the working solution on top of an AI product - part engineer, part consultant, part product manager. Posting analyses in 2026 found the title up more than 1,000 percent year on year, with the AI labs, Palantir, Databricks and Scale among the largest hirers.

How is it different from a solutions engineer or a consultant?

A forward deployed engineer builds and owns the customer's production solution on the product; a solutions or sales engineer demonstrates and scopes but usually does not build; a consultant advises and may build but does not own the product's feedback loop. The forward deployed role is closest to a founding engineer embedded with one customer at a time.

Can a fresher become a forward deployed engineer?

Rarely as a first job - most postings want shipped production software and comfort in front of customers. The fresher route is a full stack or AI engineering role with customer contact for a year or two, or a startup where you did both; switchers from consulting, solutions engineering or product management with real engineering skill are the most natural fit.

Which skills matter most?

Both halves: full stack engineering plus applied AI - model APIs, prompting and context, retrieval with permissions, structured outputs, evaluation and cost control - and the customer-facing craft of discovering the real problem, scoping, demonstrating weekly, writing clearly and handling the moment the model is wrong in front of the client.

Do I need to travel or be on site?

Often, or at least to work in the customer's time zone. The work is embedded by definition, which favours candidates near enterprise customers or willing to travel, and it is one reason the role pays well above standard engineering for the same years of experience.

How long does it take to become ready?

Count in shipped projects rather than months: a full stack application with real users, a retrieval-augmented feature over a real document set with an evaluation and a cost note, and a bounded agent with human approval - each built for a real customer or team. Nobody can promise a timeline; a case study with the customer's outcome is the evidence.

Will AI replace forward deployed engineers?

The role exists because AI products do not deploy themselves: someone has to understand the customer's problem, scope the agent to a job that pays off and bound it so it can be trusted. Gartner's forecast that more than 40 percent of agentic projects will be cancelled by 2027 is the demand signal - companies need the person who prevents that.

How do I prove forward deployed skills without the title?

A deployment for someone else: a document-answering tool for a real organisation with an evaluation set and adoption numbers, an agent that automates one workflow for a small business with approvals and logs, a case study that states the problem, the solution, the measurement and what the customer did next. Add OneRoadmap's AI Engineer certification and job simulation as verifiable evidence of the AI half.