A forward deployed engineer (FDE) is a software engineer who works directly inside a customer’s organisation to scope, build, deploy and support software, increasingly AI systems, on that customer’s own data and infrastructure. Palantir created the role. In 2026 it is a core hiring title at OpenAI, Anthropic, Databricks, Google Cloud and a growing list of Indian employers.
Key takeaways
- An FDE owns an outcome for one named customer, not a feature on a product roadmap. The job mixes engineering, deployment and consulting in roughly equal parts.
- Across 1,000 postings analysed in January 2026, Python (66%), AI agents (35%), TypeScript (35%) and AWS (32%) were the most requested skills, and 55% listed customer-facing work first.
- Verified India medians sit around ₹13–21 L a year. The ₹35–45 L entry figures quoted by some recruiters reflect a small top tier, not the market.
- Hyderabad is a top-three Indian FDE city, with Turing, Sutherland, Pulsora, EPAM, TCS and Accenture hiring locally.
- Most postings want 3–5 years of experience, so freshers and career switchers need a deliberate entry route: structured training plus a deployment they can show.
What is a forward deployed engineer?
The term borrows from the military. A forward deployed unit is stationed at the front, not at headquarters. A forward deployed engineer sits with the customer, often on site, and ships working software against that customer’s real systems, data and constraints.
Three things separate an FDE from a conventional engineer. An FDE owns a business outcome, not a ticket. An FDE writes production code on infrastructure the vendor does not control. And an FDE talks to the customer’s people every day, from the data engineer who owns the warehouse to the CFO who signs the renewal.
In practice the title covers a spectrum. At one end are Palantir-style FDEs who build custom applications on a data platform. At the other are AI-lab FDEs who integrate a foundation model into an enterprise workflow and prove it works with evals. The common thread is deployment inside the customer’s walls.
Where did the forward deployed engineer role come from?
Palantir created the role in the early 2010s. Its government and industrial customers had data problems that no packaged product could solve, so Palantir sent engineers, which it called Forward Deployed Software Engineers (FDSEs), to work in the customer’s building. According to the Pragmatic Engineer, by 2016 Palantir employed more FDEs than product engineers.
That inversion is the point. Palantir’s products, Gotham and Foundry, were shaped by what FDEs learned on site and fed back to product teams. Most AI companies have since taken the same lesson: when the technology is new and every customer’s data is messy, deployment is the product.
The model stayed niche until foundation models made it possible to build useful enterprise systems quickly, provided someone could handle the customer’s data, security review and change management. The FDE title then spread from Palantir to OpenAI, Anthropic and Databricks, and from there to Indian IT services firms, global capability centres and startups.
What does a forward deployed engineer do day to day?
Most FDE engagements run through five overlapping activities. OpenAI structures its engagements as scoping, validation with evals, then delivery, with FDEs writing code directly on customer infrastructure, as the Pragmatic Engineer piece above describes.
1. Scoping and discovery
The first week is interviews, not code. The FDE maps the customer’s workflow, finds where the data actually lives, identifies the stakeholders who can say no, and writes a scoping document with measurable success criteria.
A vague brief like “build us a support copilot” becomes “cut average handling time on tier-1 tickets by 20% within eight weeks, measured in the existing ticketing system”. Without that sentence the project cannot succeed, because nobody can tell when it has.
2. Building
In 2026 the build is usually one of three patterns: a retrieval-augmented generation (RAG) system over the customer’s documents, an agent that calls internal systems through MCP (Model Context Protocol) servers, or a batch pipeline that classifies or extracts data from enterprise records.
The stack is unglamorous and consistent. Python and FastAPI on the back end, PostgreSQL with pgvector for retrieval, a TypeScript or React front end, and the Claude or OpenAI API behind a gateway. Most of the engineering time goes into the customer’s data, not the model.
3. Deploying inside the customer’s walls
The code runs where the customer’s security team allows: a private VPC on AWS or Azure, an on-prem Kubernetes cluster, or behind a model gateway that logs every call. The FDE handles single sign-on (OIDC or SAML), role-based access, network rules and audit logging.
A system that passes the demo but fails the CISO review is not deployed. This is why enterprise integration and security are a full phase of any serious forward deployed engineer roadmap, not an afterthought.
4. Evals
Before anything reaches end users, the FDE builds an evaluation set: real inputs, expected outputs and a scoring method, often a mix of exact-match checks and an LLM acting as judge. Evals are how the FDE proves the system works, and how regressions are caught after every prompt or model change.
At OpenAI, validation with evals is a formal gate between scoping and delivery. At FDE Masters no project reaches demo day without passing its eval suite, for the same reason.
5. Customer work
Throughout the engagement the FDE runs weekly status calls, demos to executives, trains end users and handles escalations. In Bloomberry’s analysis of 1,000 FDE postings from January 2026, 55% listed customer-facing work before any technical requirement, and 68% mentioned travel.
An example from our own deployment work at Brolly Software Solutions: when we shipped Brolly VoxFlow, a spreadsheet-to-voice outbound calling system with RAG, the model integration took days. The customer’s call-list data quality, consent rules and reporting format took weeks. That ratio is normal.
Why are OpenAI, Anthropic and Databricks hiring forward deployed engineers?
Foundation models are general. Enterprise value is specific. The gap between a model API and a working system inside a bank, hospital or logistics firm is exactly the gap an FDE closes, and the labs have concluded that closing it themselves is faster than waiting for partners to learn.
- OpenAI runs a dedicated FDE team and structures each engagement as scoping, evals and delivery, with engineers writing code on customer infrastructure rather than shipping a generic integration.
- Anthropic’s FDE job description asks for engineers who ship MCP servers, sub-agents and agent skills for customers, with 4+ years of experience and a US base range of $280,000 to $320,000.
- Databricks hires Senior FDEs, including in Bengaluru, to take its data and model-serving stack into customer production.
- Google Cloud runs an Applied AI FDE function, also hiring in Bengaluru, focused on enterprise model deployments.
Want to become a Forward Deployed Engineer in Hyderabad?
Sit in on a live class, see how we scope and deploy for our training client Deccan Finance, and ask the trainer anything about the role. No sales pitch, just the work.
Forward deployed engineer vs software engineer, solutions engineer and consultant
People often ask whether an FDE is just a solutions engineer, or a consultant who codes. The short answer: an FDE combines the ownership of a software engineer, the customer exposure of a solutions engineer and the problem framing of a consultant, and is accountable for all three at once.
| Dimension | Software engineer | Solutions engineer | Consultant | Forward deployed engineer |
|---|---|---|---|---|
| Primary output | Product features | Demos, proofs of concept | Recommendations, decks | Working system in production at one customer |
| Customer contact | Rare | High, pre-sale | High | Daily, through delivery and adoption |
| Code written | Most of the week | Some, mostly throwaway | Little or none | Most of the week, production grade |
| Owns the outcome? | Owns the feature | Owns the technical win | Owns the advice | Owns the result |
| Typical home | Product team | Sales team | Advisory firm | Deployment or applied AI team |
The full comparison, including AI engineers and DevOps engineers and how to move from each role into FDE work, is in our guide to forward deployed engineer vs software engineer and related roles.
What skills does a forward deployed engineer need?
The hard skills are an applied AI stack on top of a solid software base. Bloomberry’s January 2026 review of 1,000 postings gives the clearest public picture of what employers actually list.
| Skill | Share of 1,000 FDE postings | What it means in practice |
|---|---|---|
| Python | 66% | APIs, data wrangling, async services, tests |
| TypeScript | 35% | Internal tools, dashboards, Next.js front ends |
| AI agents | 35% | Tool calling, LangGraph or Claude Agent SDK, MCP servers, human-in-the-loop controls |
| AWS | 32% | IAM, VPC, deploying into customer accounts, reading the cost bill |
| LLM engineering | 31% | Structured outputs, RAG, prompt design, cost and latency budgets |
Three skills appear in almost every posting without being named: SQL on messy data, Docker and Kubernetes for deployment, and evals. The posting says “ship production AI”. The interview tests whether you can do these three.
The soft skills are not optional. With 55% of postings putting customer-facing work first, discovery interviews, writing a scoping document, running a status call and demoing to an executive are graded skills in our programs. The complete list, with a self-assessment checklist, is in our breakdown of forward deployed engineer skills and qualifications.
What is a forward deployed engineer’s salary in India and the US?
Salary data for FDEs in India is thin and the spread is wide, so look at medians rather than headlines. The table shows the sources we consider reliable as of October 2026.
| Market | Source | Figure |
|---|---|---|
| India | Levels.fyi (12 submissions) | Median ₹17.7 L; 75th percentile ₹42.2 L |
| India | Glassdoor India | Average ₹12.9–14.2 L; 90th percentile ₹34–40 L |
| India | AmbitionBox (FDSE title) | Average ₹20.6 L |
| India, IT-services FDE practices | Public postings | ₹10–18 L |
| India, top tier (AI labs, funded startups) | CIEL HR client data, July 2026 | Entry ₹35–45 L; senior ₹70–90 L |
| US | Bloomberry, 1,000 postings | Median $173,816 |
| US, AI lab | Anthropic FDE job description | $280,000–320,000 base |
The ₹35–45 L entry figure comes from CIEL HR’s clients, which skew toward well-funded AI companies. It is real but not typical. For most Indian candidates the realistic first FDE offer is ₹10–20 L, rising quickly after one or two successful deployments. The full breakdown by city, experience and employer tier is in our forward deployed engineer salary in India guide.
Which companies hire forward deployed engineers in India and Hyderabad?
CIEL HR counted 52 organisations hiring FDEs in India in July 2026, up 130% year on year, with Bengaluru taking about half the demand and Hyderabad in the top three. CIEL puts the deployable talent pool at roughly 1,200 people, which is why the role is undersupplied.
- Bengaluru: Sarvam AI (separate 1–3, 2–5 and 4+ year tracks), Google Cloud Applied AI, Databricks, Cartesia (₹70–90 L band).
- Hyderabad: Turing, Sutherland, Pulsora, EPAM (local academy, 3,000+ India hires planned for 2026), TCS, Accenture.
- Elsewhere in India: Supervity (Mumbai), MongoDB and Coralogix (Gurgaon), Coforge’s Momentuum AI unit.
- AI labs: Anthropic opened a Bengaluru office in February 2026. OpenAI has a Delhi office, with Mumbai and Bengaluru planned.
Note the IT-services names. TCS has publicly targeted 5,900–8,900 forward-deployed AI engineers, and Accenture announced a ServiceNow FDE program in May 2026. These roles pay less than lab roles, but they are the realistic entry point for most Hyderabad candidates. We track local openings in our guide to forward deployed engineer jobs in Hyderabad.
Is forward deployed engineer a good career?
Yes, for a specific kind of engineer. Demand is real: Bloomberry measured a 1,165% year-on-year rise in FDE postings and CIEL HR a 130% rise in Indian hiring organisations. The talent pool is small. Pay at the top is among the highest in software.
The trade-offs are also real. 68% of postings mention travel. Customer deadlines do not move for your sprint. You will spend as much time in meetings and documents as in code, and your work is judged by whether the customer’s metric moved.
The role suits engineers who like owning problems end to end, can explain technical trade-offs to non-engineers, and are comfortable with ambiguity. It does not suit people who want a stable backlog and a single codebase. Since 60% of postings ask for 3–5 years of experience, freshers should plan a deliberate entry route rather than expecting a direct lab offer.
How do you become a forward deployed engineer?
There is no FDE degree. The route is a sequence of capabilities and a portfolio that proves them. In order:
- Build engineering foundations. Python with typing and tests, SQL on real enterprise data, REST APIs with authentication, Git and Linux.
- Learn cloud and delivery. One cloud (AWS is the most requested), Docker, CI/CD, Kubernetes basics and infrastructure as code.
- Learn applied AI the production way. LLM APIs with structured outputs, RAG with hybrid search and permission-aware retrieval, evals with golden datasets and CI gates, agents and MCP servers with human-in-the-loop controls.
- Learn enterprise integration and security. SSO, RBAC, ServiceNow and Salesforce integration, the OWASP LLM Top 10, prompt-injection defences, India’s DPDP Act 2023 and SOC 2 basics.
- Practise customer craft. Discovery calls, scoping documents, statements of work, status notes and executive demos.
- Ship a deployment you can show. A real system for a real user, with an eval report and a one-page case study.
- Interview with evidence, not claims. Walk through the deployment, what broke, and what the customer’s metric did.
The week-by-week version of this list, with free resources for each phase, is in our forward deployed engineer roadmap linked above. If you want the structured version, the FDE Career Program runs 18 weeks (16 teaching plus a 2-week practicum) with Monday to Friday live classes, in the classroom in Hyderabad near JNTU Metro in Kukatpally or live online. It is built for beginners and early-career engineers with any coding exposure and no AI prerequisite, and costs ₹45,000 online or ₹55,000 classroom, with EMI options available.
Engineers with 2–7 years of experience who already code should look at the 10-week FDE Advanced Program instead. Both programs end with a live two-week forward deployment, with a Brolly Group business unit or a Digital Brolly SME client as the customer, because the only way to learn deployment is to deploy.
Frequently asked questions
What is a forward deployed engineer in simple terms?
A forward deployed engineer is a software engineer who is sent to work inside a customer’s company to build and deploy a solution on the customer’s own systems. Instead of building product features for many users, an FDE builds one working system for one customer and is judged on whether it delivers the agreed business result.
Is forward deployed engineer a high paying job?
At the top, yes. US medians sit around $174,000 to $216,000 and Anthropic lists a base of $280,000 to $320,000. In India the verified median is about ₹17.7 L on Levels.fyi, with AI labs and funded startups paying ₹35 L to ₹90 L and IT-services FDE practices paying ₹10 L to ₹18 L. Pay tracks the employer tier more than the title.
Which companies have forward deployed engineers?
Palantir, OpenAI, Anthropic, Databricks, Google Cloud and Scale AI all hire FDEs. In India the list includes Sarvam AI, Cartesia, Turing, Sutherland, Pulsora, EPAM, TCS, Accenture, Coforge, Supervity, MongoDB and Coralogix. CIEL HR counted 52 Indian organisations hiring FDEs in July 2026, with Bengaluru first and Hyderabad in the top three.
Can a fresher become a forward deployed engineer?
Directly at an AI lab, rarely: only 12% of postings accept 0 to 2 years of experience. Realistic fresher routes are IT-services FDE practices, GCC academies such as EPAM in Hyderabad, startup applied AI teams and internships. A structured program that ends with a real deployment and an eval report shortens the gap considerably.
Is an FDE the same as a solutions engineer?
No. A solutions engineer supports sales with demos and proofs of concept and hands off after the deal closes. A forward deployed engineer writes production code on the customer’s infrastructure and stays until the system is live and adopted. Solutions engineers who want FDE roles usually need to add deployment, evals and security skills.
Want to become a Forward Deployed Engineer in Hyderabad?
The FDE Career Program (18 weeks, beginners welcome) and the FDE Advanced Program (10 weeks, for engineers with 2–7 years) both end with a real two-week forward deployment. Book a demo class or message us with your background and we will tell you honestly which one fits.
