What a Forward Deployed Engineer Actually Does in a Week: The Customer Shipping Cadence

Monday: The Customer Standup & Scoping War Room

The week starts not with a Jira ticket, but with a 30-minute video call where an ops manager at a logistics company explains that their warehouse management system is rejecting 12% of inbound ASNs (Advanced Shipment Notices) because the EDI 856 format has a mismatched qualifier in the N1 loop. They don’t know what an N1 loop is. They just know the warehouse floor is backing up.

Your job as a Forward Deployed Engineer is to translate business pain into an engineering scoping document before the call ends. You share your screen, open a fresh Notion page or a Linear project, and type out the acceptance criteria in plain English:

• Ingest raw EDI 856 files from the customer’s SFTP server.
• Parse the N1 loop and validate the qualifier against a lookup table provided by the 3PL.
• Flag rejections in a Slack channel with the specific file name and error reason.
• Run a manual backfill for the last 7 days.

You’re not architecting a platform. You’re defining the smallest slice of value that unblocks the warehouse by end of week. The scoping call ends with a single question: “If I ship this by Friday morning, can you test it in the afternoon?” The answer needs to be yes.

The FDE Rule of Monday: If you can’t draw the input and output on a whiteboard in 5 minutes, the scope is too big. Cut it in half.

Tuesday: The Zero-to-One Prototype Build

Tuesday is a heads-down build day. You’re not waiting for a platform team to provision a service. You spin up a Python script in a fresh repo, pull in lxml for EDI parsing, and write a 50-line function that does exactly what the whiteboard sketch promised. The code is not elegant. It has hardcoded file paths, a flat if/elif chain for qualifier codes, and a requests.post to a Slack webhook URL you grabbed from the customer’s admin.

You run it locally against the three sample files the ops manager sent on Monday. Two parse correctly. One throws a KeyError because the ISA segment uses a different delimiter. You add a 4-line fallback, rerun, and all three produce the correct Slack messages. You commit with the message “working prototype for customer review” and push to a private GitHub repo.

Tooling reality: Most FDEs live in a lightweight stack. Python or TypeScript for scripting, a cloud function or a simple container for execution, and direct API calls to the customer’s existing tools (Slack, Teams, Jira, email). You’re not building microservices. You’re building bridges.

For a deeper dive into the rapid modeling skills that make Tuesday possible, see The Highest-Leverage Skills for an FDE in the AI Era: Prompting, Data Prep, and Rapid Modeling.

Wednesday: The Integration Gauntlet

Wednesday is where the prototype meets reality. The customer’s SFTP server requires a whitelisted IP and a username/password combo that the ops manager doesn’t have. You spend 40 minutes on a call with their IT admin, who insists on a VPN tunnel instead of key-based auth. You politely push back, explain that a static IP and password will get the fix live by Friday, and offer to rotate credentials weekly. They agree.

You update the script to use paramiko for SFTP, add environment variables for credentials, and deploy it as a scheduled Cloud Run job that fires every 15 minutes. The first run in production pulls 200 files, processes them in 12 seconds, and posts 24 Slack messages for the rejected ASNs. The ops manager confirms the list matches their manual audit. You’ve just saved someone 10 hours a week of spreadsheet reconciliation.

The FDE Rule of Wednesday: The integration problem is never the technology. It’s authentication, network policy, and the person who holds the credentials. Find that person early.

Thursday: Hardening, Testing, and the Silent Demo

Thursday morning, you add error handling for the three edge cases that appeared in the production run: empty files, files with non-EDI content mistakenly dropped on the SFTP, and a 60-second timeout on large batches. You wrap the main loop in a try/except that logs failures to a shared Google Sheet the ops team already uses. No new dashboards. No new tools.

You write a 1-page runbook in the customer’s Notion workspace:

• What the script does.
• How to restart it.
• Who to call if it breaks (you, for the first 2 weeks).

You then run a “silent demo”: you don’t present. You just send the Slack channel a summary of the last 24 hours of processing, ask the ops team to spot-check 5 files, and wait for their thumbs up. They respond in 15 minutes with “looks good.” You’ve passed the silent demo. The real demo is the work doing itself.

If you’re preparing for an FDE interview, this pattern of building-then-hardening is exactly what loops test for. See The FDE Interview Loop: How to Prepare as a Builder, Not a Talker.

Friday: Shipping to Production & the Handoff

Friday morning, you flip the Cloud Run job from “manual trigger” to “scheduled every 15 minutes.” You add a simple monitoring heartbeat: if the job doesn’t complete in 5 minutes, a Dead Letter Queue alert pings you via PagerDuty. You’ll be on call for the weekend, then transition to the customer’s internal team.

You spend 30 minutes on a handoff call with a junior engineer on the customer’s side. You walk them through the repo, the runbook, and the alerting setup. You make one thing clear: “This is not a product. It’s a point solution. If the EDI format changes, you’ll need to update the parser. Here’s the section of code to change.”

The week ends with a shipped solution that’s processing files in production, a customer who can measure the impact (rejected ASNs down from 12% to 0%), and a clean handoff. Next Monday, you’ll be on a different customer call, in a different industry, solving a different problem.

The FDE Weekly Rhythm Cheat Sheet

This rhythm repeats weekly, but the domain changes. One week it’s EDI parsing for logistics. The next it’s a computer vision model for a manufacturing QA line. The week after, it’s a RAG pipeline over a legal firm’s contract repository. The constant is the cadence: scope Monday, build Tuesday, integrate Wednesday, harden Thursday, ship Friday.

For a concrete example of the kind of RAG pipeline an FDE might build in a week, see Build a Discord Community FAQ Bot with RAG on Qdrant Free Tier.

Comp and Career Context: Is This Rhythm Worth It?

The weekly rhythm is intense. You’re context-switching between industries, codebases, and customer personalities every 5 days. Burnout is real if you don’t set boundaries. But the comp and career trajectory reflect the intensity.

Forward Deployed Engineer compensation at top-tier firms (Palantir, Scale AI, Applied Intuition) typically ranges from $150K to $250K base, with total compensation including equity often landing between $200K and $350K for mid-to-senior IC roles. At consulting-heavy shops like Deloitte, FDE roles skew lower on base ($120K–$160K) but offer faster paths to client-facing leadership. The premium exists because the role demands a rare combination: the technical depth to ship production code, the communication skills to run customer calls solo, and the judgment to scope problems without a PM buffer.

Is it worth it? If you optimize for variety, autonomy, and direct customer impact, yes. If you prefer deep, uninterrupted focus on a single codebase or system, a pure software engineering role will fit better. FDEs are real engineers—they write code, design systems, and debug production issues—but they do it inside the customer’s context, not behind an internal API boundary.

FAQ: Forward Deployed Engineer Weekly Rhythm

What is the average salary of a forward-deployed engineer? At product-focused companies, base salaries typically range from $150K–$250K, with total comp reaching $200K–$350K. Consulting-adjacent roles may start lower ($120K–$160K). Equity and deployment bonuses can significantly shift the top end.

Is being a forward-deployed engineer worth it? Worth it if you value rapid learning, customer exposure, and shipping weekly. Less worth it if you want deep specialization in one technical domain or strict work-life predictability.

Are forward-deployed engineers real engineers? Yes. FDEs write production code, design data models, debug distributed systems, and handle on-call rotations. The difference is the deployment context: they work inside the customer’s environment rather than on the core product.

How much does a forward deployed engineer make at Deloitte? Deloitte and similar consulting firms typically offer $120K–$160K base for FDE-equivalent roles, with bonuses tied to utilization and project delivery. The upside is faster promotion to engagement management and partner-track roles.

What skills matter most for this weekly rhythm? Rapid prototyping in Python or TypeScript, strong debugging instincts, comfort with authentication and network configuration, and the ability to write clear, non-technical acceptance criteria. For a full breakdown, see The Highest-Leverage Skills for an FDE in the AI Era.