# What is a Forward Deployed Engineer? Author: Sachin Bhatnagar | Published: 2026-11-05 | Canonical: https://stacksauce.dev/blog/what-is-a-forward-deployed-engineer ## **What Is a Forward Deployed Engineer, and Why Is Everyone Suddenly Hiring Them?** Every few years, the tech industry invents a new job title. Then it acts surprised when nobody understands it. A few years ago it was "prompt engineer". This year, the title I keep seeing everywhere is "Forward Deployed Engineer", or FDE for short. The numbers are hard to ignore. A job report by Rung, updated on 28 October 2026, counted 396 open FDE roles across 13 companies. Databricks alone had 117 of them. Palantir had 106, Salesforce had 96 and OpenAI had 35. A separate analysis by a firm called Bloomberry, quoted by the recruiting company Paraform, found that FDE roles grew 1,165% in one year. So I am starting a short series on this role. This is Part 1. Today we will answer the simple questions first. What is an FDE? Why do companies want them right now? And should you, as a beginner, even care? ### **First, what does "forward deployed" mean?** The phrase is borrowed from the army. A "forward deployed" unit is placed close to the action, far from headquarters. In tech, it means an engineer who works close to the customer. Most software engineers sit inside their own company and build the product. An FDE goes to the customer's side and makes the product work for that customer's real problem. Sometimes this means flying to their office. Most days it means long calls, shared screens and a lot of patience. Paraform describes the job in one neat line. An FDE "writes production code", but writes it "inside a customer's environment." Production code simply means code that real people use for real work every day. It is the opposite of practice code that lives on your laptop. Here is the easiest way I can explain it. Think of a shop that sells ready-made suits. The factory makes the suit in standard sizes. But very few people are a perfect standard size. So the shop sends a tailor to your home. The tailor measures you, adjusts the sleeves, fixes the waist and makes the suit fit your body. The product team is the factory. The FDE is the tailor. ### **Why now? Blame AI, or thank it** AI companies have a funny problem. Their tools look magical in a demo. Then they enter a real company, and real companies are messy. The data sits in ten different systems. Some software from 2009 refuses to talk to anything built after 2015. The security team has a hundred questions. The staff do not trust the new tool, and some of them are quietly hoping it fails. In a sales pitch, an AI agent solves your problem in seven minutes. In real life, the same agent waits seven days for someone in IT to approve one password. Somebody has to sit in the middle of all this and make the thing actually work. That somebody is the FDE. Look again at the companies hiring: Databricks, Palantir, Salesforce, OpenAI, Scale AI and Cohere. All of them sell AI or data platforms to large businesses. Every big deal they sign needs people who can connect the shiny product to the customer's untidy reality. Palantir made this role famous long before the current AI boom. Now everyone else has copied the idea. ### **What does an FDE actually do all day?** Rung studied 300 full FDE job descriptions. In 99% of them, the role was customer-facing. That means you talk to the customer regularly. About 80% mentioned travel or on-site work. Let me give you a made-up but realistic example. Imagine an insurance company buys an AI tool. They want it to read thousands of claim forms and flag the suspicious ones. Here is what the FDE might do over a few weeks: - Sit with the claims team and understand how they work today. - Look at their data and find out where the forms are stored, in what format, and how messy they are. - Build a prototype. A prototype is a small, rough working version that shows the idea works. - Test it on real forms and fix everything that breaks. Something always breaks. - Train the staff to use it, and answer the same question patiently for the fifth time. - Go back to their own product team and say, "Three customers asked for this feature. Please build it." That last point matters a lot. The FDE is the company's ears. They bring real customer pain back to the people who build the product. ### **The skills, in plain words** Here is what those 300 job descriptions asked for, according to Rung: - **Python (79%).** Python is a programming language known for being easy to read. It is the most common language in AI work. - **APIs and integrations (66%).** An API is a set of rules that lets two pieces of software talk to each other. Integration means connecting your product to the customer's existing systems. - **Cloud (61%).** Cloud means renting computers over the internet from companies like AWS, Google Cloud or Microsoft Azure. - **JavaScript, React or TypeScript (59%).** These are the main tools for building websites and web apps. - **Data, SQL and pipelines (52%).** SQL is a language for asking questions to a database. A pipeline is a series of steps that moves data from one place to another and cleans it on the way. - **LLMs and GenAI (31%).** An LLM, or large language model, is an AI system trained on huge amounts of text. ChatGPT and Claude are built on LLMs. ![*Bar chart of the skills asked for in 300 Forward Deployed Engineer job posts, based on Rung's October 2026 report. Customer-facing 99%, travel or on-site 80%, Python 79%, APIs and integrations 66%, cloud 61%, JavaScript, React and TypeScript 59%, data, SQL and pipelines 52%, and LLMs and GenAI 31%, highlighted in red.*]() _*Bar chart of the skills asked for in 300 Forward Deployed Engineer job posts, based on Rung's October 2026 report. Customer-facing 99%, travel or on-site 80%, Python 79%, APIs and integrations 66%, cloud 61%, JavaScript, React and TypeScript 59%, data, SQL and pipelines 52%, and LLMs and GenAI 31%, highlighted in red.*_ Look closely at that last number. Only 31% of postings mention LLMs. Now most will say, "FDE is an AI job, so I must learn AI first." Partly agreed. AI knowledge helps, and the AI labs do want it. But look at what sits at the top of the list. Python, APIs, cloud and SQL. These are solid, old-fashioned engineering skills. The AI part sits on top of them. Then come the human skills, and in my view these decide who succeeds. Rung says the interviews include "an open-ended customer case". In simple words, they give you a messy business problem and watch how you think. Can you ask good questions? Can you explain a technical idea to someone who is not technical? Can you stay calm when the demo crashes in front of the client? I have spent more than 20 years teaching people technology. I can tell you that explaining things clearly is a skill you can practise. Nobody is simply born with it. One reason why I became a teacher is because a lot of my own teachers were terrible at simplifying concepts. In my head I would break them down easily but I could see others struggle. As a teacher, explaining things became a skill. ### **What does it pay?** This is the part everyone scrolls down to first, so let me be careful. Rung found 26 US roles that shared a pay range. The typical range ran from $185,000 to $300,000 a year. Paraform puts the typical base salary at $150,000 to $217,000. These are US numbers for US jobs. Now the honest part for my Indian readers. Out of Rung's 396 roles, only 8 were based in India. The United States had 169. The United Kingdom had 41. So please do not convert dollars to rupees and start booking the Fortuner test drive just yet. But here is something interesting. Indian IT companies are already moving their freshers towards this kind of work, even if they do not use the title. On 15 October 2026, Business Today reported that Infosys plans to hire 20,000 freshers and will train them in AI before putting them on client projects. The work described includes helping clients use AI to write, test and maintain software, and upgrading old systems. That is forward deployed work under a different name. Honestly, Indian IT services has done a version of this for decades. We just called it "onsite". The difference now is that the product being deployed is AI, and the engineer is expected to build, debug and explain, all in the same week. ### **Is this role right for you?** You will probably enjoy FDE work if: - You like solving other people's problems more than polishing your own code. - You enjoy talking to people, or at least you do not run away from it. - You can handle messy, unclear situations where nobody gives you a neat task list. - You like seeing your code used by real people within days. You may prefer a regular software role if you want long, quiet stretches of deep work on one product. There is no shame in that. Both kinds of engineers are needed, and the world would be a strange place if everyone wanted to be on client calls all day. ### **What you can do this week** You do not need to be job-ready tomorrow. But you can start building the right habits now. 1. **Get comfortable with Python and SQL basics.** You do not need to master them yet. Learn enough to read data, clean it and answer simple questions from it. 2. **Build one small project for a real person.** Pick a shopkeeper, your college office or a relative's small business. Talk to them first. Ask what wastes their time. Then build something small that helps. 3. **Practise the two-minute explanation.** Explain your project to a friend who does not code. If they look confused, the problem is your explanation. 4. **Read ten FDE job posts.** Search for "Forward Deployed Engineer" on any job site. Write down the words that keep repeating. That list is your study plan.