How to switch into Forward Deployed Engineering
To switch into Forward Deployed Engineering, keep the engineering you already have and add what most FDE postings ask for but most engineering jobs never train: customer-facing work (87% of open roles ask for it), Python (63%) and APIs and system integration (54%), then prove it with one deployed project. This page shows what transfers from your current role, what to learn and which open roles fit, from 454 live FDE job postings at 53 companies.
Job data updated: 12 Oct 2026
Plan your switch
Readiness against real postings
30 out of 100
For every open FDE role at your level, the share of the skills it asks for that you ticked, averaged across those roles.
Your top 5 gaps, by demand
- Customer-facing work and communication 84% of open roles at your level ask for it
- LLMs, agents and RAG 53% of open roles at your level ask for it
- Travel or on-site work with customers 43% of open roles at your level ask for it
- Security and compliance 43% of open roles at your level ask for it
- TypeScript, JavaScript and front end 38% of open roles at your level ask for it
- Python 58% of open roles at your level ask for it
- APIs and system integration 47% of open roles at your level ask for it
- Java, Go, C++ or another systems language 27% of open roles at your level ask for it
- Distributed systems and system design 29% of open roles at your level ask for it
- Machine learning 25% of open roles at your level ask for it
- Data pipelines and ETL 23% of open roles at your level ask for it
- Cloud (AWS, GCP, Azure) 22% of open roles at your level ask for it
- SQL and relational databases 21% of open roles at your level ask for it
- Evals and AI quality measurement 19% of open roles at your level ask for it
- Docker, Kubernetes and infrastructure as code 10% of open roles at your level ask for it
No gaps left in our skill list. Go straight to the portfolio and interview weeks.
Your 8-week plan
- Week 1
Customer-facing work and communication
Interview three people about a workflow they dislike and write up the real problem, not the feature they asked for.
Read: The problem the client brings is rarely the real one: how FDEs clarify requirements before writing code · One deployment, four audiences: how FDEs work with a client's engineers, business teams and executives
Watch: How to communicate technical knowledge to diverse audiences (without coming across as a know-it-all)
Roadmap stage: Customers - Week 2
LLMs, agents and RAG
Build an agent loop with tool calls over your own data, without a framework, and log every step.
Read: LLM basics for FDEs: tokens, context windows, temperature and why models make things up · Build an agent loop in Python without a framework: one tool, one for loop, four common mistakes
Watch: Lessons learned in bringing a RAG chatbot with access to 50k+ diverse documents to production
Roadmap stage: Applied AI - Week 3
Security and compliance
Fill in a security questionnaire for your own project: data flows, personal data, access control.
Read: A client sends a security questionnaire: how FDEs use SOC 2 and ISO 27001 to get through review · Masking personal data before it reaches an LLM: under Vietnam's new law, swapping names for labels may not be enough
Watch: Operationalizing AI Agents in regulated industries for scale and safety
Roadmap stage: Deployment - Week 4
TypeScript, JavaScript and front end
Build a one-page React and TypeScript demo on a public API, with loading and error states.
Read: Six steps to a client React demo, starting from a machine without Node · Writing the API call layer for a TypeScript agent: timeouts, retries and idempotency keys
Watch: Lessons learned in bringing a RAG chatbot with access to 50k+ diverse documents to production
Roadmap stage: Broad engineering - Week 5
Machine learning
Train a baseline model on tabular data and write down when you would not use deep learning.
Read: Classical ML or deep learning: choose by the shape of the client's data, not by fashion · Reading "Designing Machine Learning Systems" as an FDE: start with chapters 3, 4, 9 and 10
Watch: Lessons learned in bringing a RAG chatbot with access to 50k+ diverse documents to production
Roadmap stage: Applied AI - Week 6
Ship a portfolio project
Ship one project end to end with a README, an eval set and a short write-up of the decisions you made.
Read: Building an FDE portfolio repo: a simple agent, evals built from real failures, a README and a 3-minute video · Anthropic and Palantir’s FDE job posts don’t ask for AI certificates. They ask what you have shipped
Roadmap stage: Career - Week 7
Practise the interview
Practise the decomposition round out loud: clarify the goal for the first 5–15 minutes before proposing a solution.
Read: Algorithm practice prepares you for only one round of the FDE interview loop · Telling project stories in FDE interviews: how to answer "What trade-offs did you make?"
Watch: I Left Google for FDE Role — Here's How My Salary, Work & Life Actually Changed
Roadmap stage: Career
The FDE interview → - Week 8
Apply
Rewrite your CV around measured results, then apply to the open roles that fit below.
Read: The FDE CV: drop the framework list, state deployment results with a baseline · What FDE recruiters look for: code gets you in, customer communication sets the ranking
Roadmap stage: Career
See all open FDE roles → - Week 0
Python
Write a small command-line tool that pulls data from an API, cleans it and loads it into a database, with tests.
Read: Python and SQL for FDEs: count the keys and check the JOIN before trusting the numbers · Making 20,000 LLM requests in Python without 429 failures or hung scripts
Watch: Lesson 3: Pipeline Builder & Transformations | Palantir Hands-On Bootcamp
Roadmap stage: Broad engineering - Week 0
APIs and system integration
Connect two APIs end to end with retries, idempotency keys and a webhook receiver.
Read: APIs and webhooks for FDEs: build every integration as if everything arrives twice · When the ERP has no decent API: how FDEs integrate without inheriting the old system
Watch: Lesson 3: Pipeline Builder & Transformations | Palantir Hands-On Bootcamp
Roadmap stage: Broad engineering - Week 0
Java, Go, C++ or another systems language
Read an unfamiliar Java or Go codebase and trace one request from end to end.
Read: Reading a customer's Java, Go or Node.js codebase when you only know Python
Watch: Lesson 3: Pipeline Builder & Transformations | Palantir Hands-On Bootcamp
Roadmap stage: Broad engineering - Week 0
Distributed systems and system design
Load-test your service, find the bottleneck and write the numbers down.
Read: Latency, throughput, scalability: answering the client's infrastructure team with measurements · Alex Xu's System Design Interview: keep the four-step framework, add the AI layer yourself
Watch: Operationalizing AI Agents in regulated industries for scale and safety
Roadmap stage: Broad engineering - Week 0
Data pipelines and ETL
Build an incremental pipeline that can rerun safely after a failure without duplicating rows.
Read: Hands-on: an incremental pipeline you can rerun without duplicating data · dbt: when client-site SQL needs git, tests and history too
Watch: Lesson 3: Pipeline Builder & Transformations | Palantir Hands-On Bootcamp
Roadmap stage: Deployment - Week 0
Cloud (AWS, GCP, Azure)
Deploy one small service to a cloud account with least-privilege access, then list what would change inside a customer's VPC.
Read: On-prem, VPC or SaaS: FDEs need a map of what runs where · Bedrock, Azure OpenAI, Vertex AI: getting the model into the customer's cloud
Watch: Operationalizing AI Agents in regulated industries for scale and safety
Roadmap stage: Deployment - Week 0
SQL and relational databases
Take a real dataset, write the five questions a customer asks on day one as SQL, and check the row counts of every join.
Read: First time in a customer's database: read the schema, query a replica, lock the session to read-only · Python and SQL for FDEs: count the keys and check the JOIN before trusting the numbers
Watch: Lesson 3: Pipeline Builder & Transformations | Palantir Hands-On Bootcamp
Roadmap stage: Broad engineering - Week 0
Evals and AI quality measurement
Write a 50-case eval set for your LLM project and run it on every change.
Read: Hands-on: building a 50-case eval for a Q&A assistant in Python · Writing your first eval with Inspect AI: dataset, solver, scorer and reading the log
Watch: The agent evaluation revolution
Roadmap stage: Measurement - Week 0
Docker, Kubernetes and infrastructure as code
Put your project in a container, run it on a local Kubernetes cluster and break it on purpose to practise debugging.
Read: Diagnosing CrashLoopBackOff, OOMKilled and DNS failures on a customer's Kubernetes cluster with kubectl · Docker for FDEs: when the image reaches the customer's server on a USB stick
Watch: Operationalizing AI Agents in regulated industries for scale and safety
Roadmap stage: Deployment - Week 0
Make a demo production-ready
Take a demo and add error handling, tests, monitoring and a runbook.
Read: From POC to production: why so many AI projects stop at the demo · From solutions engineer to FDE: rebuilding a PoC to production standard
Roadmap stage: Deployment - Week 0
Practise discovery
Sit with a user and map how the work is done today before proposing anything.
Read: A day beside the user: how FDEs watch real work to find what clients never mention · A practical guide: map the client's real process before letting AI automate it
Watch: How to communicate technical knowledge to diverse audiences (without coming across as a know-it-all)
Roadmap stage: Customers - Week 0
Write for the customer
Write the design doc and the weekly status report a customer would read.
Read: The weekly client report: one page, three questions and a decisions section that unblocks the project · Write a design doc the client's engineers approve before you code
Roadmap stage: Customers
Open roles that fit
At your level and region, best fit first: the share of each posting's skill asks you already cover.
- Hardware Engineer - Forward Deployed Saronic · North America you have 3 of 3 skills it asks for
- Hardware Engineer - Forward Deployed Saronic · North America you have 3 of 3 skills it asks for
- Software Engineer - Forward Deployed Saronic · North America you have 4 of 5 skills it asks for
- Software Engineer - Forward Deployed Saronic · North America you have 4 of 5 skills it asks for
- Forward Deployed Strategist Chalk · North America you have 3 of 4 skills it asks for
- Forward Deployed Strategist Chalk · North America you have 3 of 4 skills it asks for
- Forward Deployed Engineer - Software Engineer - Mexico ElevenLabs · Other you have 2 of 3 skills it asks for
- Forward Deployed Engineer - Software Engineer - Argentina ElevenLabs · Other you have 2 of 3 skills it asks for
No open roles match this level and region right now. Try "Anywhere".
What open FDE roles ask for
We read the requirements of all 454 open Forward Deployed Engineer postings and count how many ask for each skill. The table is what employers ask for right now, split by the seniority in the job title.
Databricks (124) and Palantir (109) post 51% of the open roles between them, so their wording weighs on these shares. Early-career roles are few (45), so their column moves a lot when one company posts.
By starting role
Pick the job you have now. Each card shows what carries over, the gaps ordered by how many open roles ask for them, and what to read first. Ticks in the planner start from the skills listed under "what transfers".
Backend engineer
What transfers
Python · SQL and relational databases · APIs and system integration · Java, Go, C++ or another systems language · Distributed systems and system design
Production code, APIs, databases and on-call habits: the base every FDE posting assumes. 63% of open roles ask for Python, 54% for APIs and integration and 41% for distributed systems or system design.
Your gaps, by demand
What a backend job rarely trains is the customer side. Our analysis of the switch finds many software engineers lack direct customer experience, the habit of uncovering undocumented workflows and experience taking LLMs to production.
- Customer-facing work and communication 87%
- Travel or on-site work with customers 46%
- TypeScript, JavaScript and front end 46%
- LLMs, agents and RAG 45%
- Machine learning 37%
- Data pipelines and ETL 36%
Learn first
- The problem the client brings is rarely the real one: how FDEs clarify requirements before writing code
- Six steps to a client React demo, starting from a machine without Node
- LLM basics for FDEs: tokens, context windows, temperature and why models make things up
- Classical ML or deep learning: choose by the shape of the client's data, not by fashion
Read first
Full-stack or front-end engineer
What transfers
TypeScript, JavaScript and front end · APIs and system integration · SQL and relational databases
Putting something a customer can click on in front of them within days: 46% of open roles name TypeScript, JavaScript or front-end work, and a working demo is how an FDE turns a vague request into a concrete one.
Your gaps, by demand
The gaps are usually Python (63% of roles), data work and LLM systems, plus owning a deployment inside someone else's infrastructure rather than your own.
- Customer-facing work and communication 87%
- Python 63%
- Travel or on-site work with customers 46%
- LLMs, agents and RAG 45%
- Java, Go, C++ or another systems language 43%
- Distributed systems and system design 41%
Learn first
- The problem the client brings is rarely the real one: how FDEs clarify requirements before writing code
- Python and SQL for FDEs: count the keys and check the JOIN before trusting the numbers
- LLM basics for FDEs: tokens, context windows, temperature and why models make things up
- Reading a customer's Java, Go or Node.js codebase when you only know Python
Read first
Data engineer
What transfers
Python · SQL and relational databases · Data pipelines and ETL · Cloud (AWS, GCP, Azure)
Getting messy customer data into shape is a large part of the job: our guide calls customer data onboarding core FDE work. 36% of open roles ask for data pipelines or ETL.
Your gaps, by demand
The gaps are the customer conversation, LLM applications and evals, and building the application your pipeline feeds, not just the pipeline.
- Customer-facing work and communication 87%
- APIs and system integration 54%
- Travel or on-site work with customers 46%
- TypeScript, JavaScript and front end 46%
- LLMs, agents and RAG 45%
- Java, Go, C++ or another systems language 43%
Learn first
- The problem the client brings is rarely the real one: how FDEs clarify requirements before writing code
- APIs and webhooks for FDEs: build every integration as if everything arrives twice
- Six steps to a client React demo, starting from a machine without Node
- LLM basics for FDEs: tokens, context windows, temperature and why models make things up
Read first
ML or AI engineer
What transfers
Python · Machine learning · LLMs, agents and RAG · Evals and AI quality measurement
Our guide for data scientists and ML engineers calls ML experience and a debugging mindset genuine advantages: 37% of open roles ask for machine learning and 45% for LLMs, agents or RAG.
Your gaps, by demand
The same guide puts the gap after the model: problem decomposition, production code and end-to-end ownership, with a specific customer deciding what counts as done.
- Customer-facing work and communication 87%
- APIs and system integration 54%
- Travel or on-site work with customers 46%
- TypeScript, JavaScript and front end 46%
- Java, Go, C++ or another systems language 43%
- Distributed systems and system design 41%
Learn first
- The problem the client brings is rarely the real one: how FDEs clarify requirements before writing code
- APIs and webhooks for FDEs: build every integration as if everything arrives twice
- Six steps to a client React demo, starting from a machine without Node
- Reading a customer's Java, Go or Node.js codebase when you only know Python
Read first
DevOps, SRE or platform engineer
What transfers
Docker, Kubernetes and infrastructure as code · Cloud (AWS, GCP, Azure) · Security and compliance · Distributed systems and system design
Deploying into environments you do not control (customer VPCs, on-prem and air-gapped servers) and handling incidents. 35% of open roles ask for cloud, 33% for security or compliance and 10% for Docker, Kubernetes or infrastructure as code.
Your gaps, by demand
The gaps are application code in Python, LLM systems and leading the conversation with the customer instead of answering tickets.
- Customer-facing work and communication 87%
- Python 63%
- APIs and system integration 54%
- Travel or on-site work with customers 46%
- TypeScript, JavaScript and front end 46%
- LLMs, agents and RAG 45%
Learn first
- The problem the client brings is rarely the real one: how FDEs clarify requirements before writing code
- Python and SQL for FDEs: count the keys and check the JOIN before trusting the numbers
- APIs and webhooks for FDEs: build every integration as if everything arrives twice
- Six steps to a client React demo, starting from a machine without Node
Read first
Solutions or sales engineer
What transfers
Customer-facing work and communication · Travel or on-site work with customers · APIs and system integration
Customer conversations, demos and travel, which most engineers have to learn from scratch: 87% of open roles ask for customer-facing work and 46% mention travel or on-site work.
Your gaps, by demand
Our guide to the move names the gap: from demo code to production code, and taking responsibility for how the deployment turns out.
- Python 63%
- TypeScript, JavaScript and front end 46%
- LLMs, agents and RAG 45%
- Java, Go, C++ or another systems language 43%
- Distributed systems and system design 41%
- Machine learning 37%
Learn first
- Python and SQL for FDEs: count the keys and check the JOIN before trusting the numbers
- Six steps to a client React demo, starting from a machine without Node
- LLM basics for FDEs: tokens, context windows, temperature and why models make things up
- Reading a customer's Java, Go or Node.js codebase when you only know Python
Read first
Consultant or analyst
What transfers
Customer-facing work and communication · SQL and relational databases · Travel or on-site work with customers
Discovery, stakeholder management and writing for executives transfer directly: 87% of open roles ask for customer-facing work and communication.
Your gaps, by demand
The gap is engineering. An FDE writes the production code and owns the outcome, so Python (63%), APIs (54%) and shipping a working system come first.
- Python 63%
- APIs and system integration 54%
- TypeScript, JavaScript and front end 46%
- LLMs, agents and RAG 45%
- Java, Go, C++ or another systems language 43%
- Distributed systems and system design 41%
Learn first
- Python and SQL for FDEs: count the keys and check the JOIN before trusting the numbers
- APIs and webhooks for FDEs: build every integration as if everything arrives twice
- Six steps to a client React demo, starting from a machine without Node
- LLM basics for FDEs: tokens, context windows, temperature and why models make things up
Read first
New graduate
What transfers
Python
Fresh fundamentals count: early-career postings ask for Python (87%), TypeScript or JavaScript (80%) and Java, Go or C++ (80%) more than anything else.
Your gaps, by demand
The door is narrow: 45 of 454 open roles (10%) are early-career by title. Our analysis of new-grad FDE hiring finds those roles tied to a fixed graduation window, and for most engineers in Vietnam a more realistic path of 12–18 months of deployment experience first.
- Customer-facing work and communication 87%
- APIs and system integration 54%
- Travel or on-site work with customers 46%
- TypeScript, JavaScript and front end 46%
- LLMs, agents and RAG 45%
- Java, Go, C++ or another systems language 43%
Learn first
- The problem the client brings is rarely the real one: how FDEs clarify requirements before writing code
- APIs and webhooks for FDEs: build every integration as if everything arrives twice
- Six steps to a client React demo, starting from a machine without Node
- LLM basics for FDEs: tokens, context windows, temperature and why models make things up
Read first
Frequently asked questions
Can a backend engineer become an FDE?
Yes. Backend skills are what most postings ask for first: 63% of the 454 open FDE roles ask for Python, 54% for APIs and integration and 41% for distributed systems or system design. What backend jobs rarely train is the customer side: 87% ask for customer-facing work. Our analysis of the switch finds many software engineers lack direct customer experience, uncovering undocumented workflows and taking LLMs to production.
Do I need customer-facing experience?
For most roles, yes: 87% of open FDE postings ask for customer-facing work, stakeholder management or communication with executives, more than any technical skill. Our analysis notes that Anthropic's posting asks for 4+ years in a technical, customer-facing role, and suggests building that experience in your current job first.
How long does it take to switch into an FDE role?
It depends on the gap. The planner on this page gives eight weeks to work through your five biggest skill gaps, ship one project and practise the interview; that is preparation, not a hiring timeline. For engineers without deployment experience, our analysis of new-grad FDE hiring says the more realistic path for most engineers in Vietnam is 12–18 months of deployment experience before moving into an FDE role. Only 45 of 454 open roles are early-career by title.
Do FDEs travel?
Often. 46% of open FDE postings mention travel or on-site work with customers, rising to 61% of senior roles. Our analysis of travel requirements puts the usual range at 25–50% of time at the customer, and our look at remote FDE jobs found even remote-friendly Cohere asking for 20–40% travel. Ask for the actual figure before accepting an offer.
Is FDE a step down from software engineering?
Not on the evidence we have. US FDE postings that publish a range have a median base salary of $216k (180 posts, see FDE salary data); we do not collect software engineer pay, so we cannot compare the two directly. The role has its own ladder: our salary ladder analysis found an OpenAI FDE manager tier asking for 8+ years of experience, and our guide to career paths lists principal FDE, product manager, engineering manager and founder as routes out. What changes is how you are judged: on one customer's outcome rather than on a component you own.
What do FDE interviews test?
According to the course our guide to the decomposition round draws on, the loop has four rounds: decomposition, customer simulation, technical and behavioural. The decomposition round is scored on structured decomposition, empathy for the end user, reasoning about data and judgement on scope; the most common mistake is jumping straight to a solution. See the FDE interview, round by round.
Do I need AI or LLM experience to become an FDE?
Increasingly: 45% of open FDE postings ask for LLMs, agents or RAG and 14% mention evals or evaluation, while 37% ask for machine learning more broadly. If you have none, the plan above puts LLMs and evals in your first weeks when they are among your biggest gaps, starting from LLM basics.
Which skills should I learn first?
The ones most open roles ask for that you lack. Across all 454 postings the most requested are customer-facing work and communication (87%), python (63%), APIs and system integration (54%), travel or on-site work with customers (46%), typeScript, JavaScript and front end (46%). Tick what you already have in the planner and it orders the rest by demand at your level.
How this works
Data. 454 open Forward Deployed Engineer roles at 53 companies, read from official careers pages (Greenhouse, Ashby, Lever) and refreshed several times a day; the page rebuilds when the roles change.
Skills. A fixed list of 15 skills, each matched by one fixed pattern in the requirements part of the posting. Benefits, pay, "about us" and equal-opportunity sections are dropped first, and so is the company's own name, so a company describing itself or its product does not count as a skill it asks for. A posting either asks for a skill or not; how strongly it asks is not measured.
Seniority. Read from the job title: intern, graduate, junior and associate are early career; senior, staff, principal, lead and manager are senior; the rest are mid level. With 5+ years you are matched to senior and mid-level roles.
Readiness. For every open role at your level, we take the skills it asks for and the share of them you ticked, then average across those roles. 100 means you cover every listed skill of every role at your level; it does not measure how good you are at them. Roles in which we can read no skill are left out.
What it cannot see. Domain knowledge, security clearance, languages and degree requirements are not in the list. A posting that says "Python" in a list of examples counts the same as one that requires it.
What is an FDE? →The 8-stage roadmap →The FDE interview →Hiring data →
Suggested citation: FDE Times, “How to switch into Forward Deployed Engineering”, https://fdetimes.net/en/switch-to-fde/, 12 Oct 2026.