From engineer or domain expert to forward deployed engineer.
A program for people who will sit with a client, understand how the work really happens, propose a solution and deliver a working proof of concept with LLMs such as Claude. It takes people from either a software or a domain and business background and brings both to the same finish line.
Before module 01. Read the basics: what a model, an LLM, context and a harness are, Claude app versus API, open versus closed models, VPCs, and data privacy versus security. Then set up GitHub, VS Code and an AI coding assistant, and start your learning journal.
AI runs on six layers. Each layer buys from the one below it and sells to the one above. Most of the money goes into the bottom layers; most of what makes one company's AI different from another's is built in the top two. That is where the FDE works: choosing a model from the layer below, then building the agent and observability layers so the client gets AI that fits, is safe, costs as little as possible, can be maintained and does what it was set up to do.
Observabilitywatch, evaluate, governFDE builds here
Records every step the system takes, scores the quality of its answers, catches problems before users do, and keeps the audit trail a regulator or an unhappy customer will ask for. Small layer, big consequences.
The logic around the model: routing, tools and their permissions, memory, retrieval, approvals and the loop that turns a model into something that does real work. Where the client's solution actually lives.
Claude Agent SDK, OpenAI Agents SDK, Google ADK, LangGraph, Microsoft Agent Framework, MCP
Modelweights and inferenceFDE chooses here
The models themselves, plus the serving that turns them into answers at a price per token. Changes fastest and gets cheaper fastest.
Anthropic, OpenAI, Google, Meta, Mistral, DeepSeek, Qwen; served by them or by clouds and hosts
Computewho runs the machines
Buys the chips, builds the data centers and rents capacity by the hour, or runs it privately. Where most of the money is spent.
AWS, Microsoft Azure, Google Cloud, Oracle Cloud, CoreWeave, your own data center
Siliconwho makes the chips
Designs and manufactures the processors, memory and networking that everything else runs on. The hardest layer to enter.
NVIDIA, AMD, TSMC, Samsung, SK Hynix, Micron; in-house chips such as Google TPU and AWS Trainium
Energypower, grid, cooling
Electricity, transmission and cooling for data centers. Now a real limit on how fast the rest can grow.
Utilities and grid operators, power generators
What the FDE does across those layers
Find the real jobThe process, the bottleneck, the measure
Choose a suitable modelModel layer: right size for each step
Build the agent layerCode, tools, data access, human approval
Secure itLeast privilege, PII, injection, audit
Keep cost downSmaller models, caching, batching
Observe and evaluateTraces, evals, alerts, drift
Hand over to maintainRunbook, owner, support model
The FDE does not train models or run data centers. They pick the right model and build, secure, measure and hand over the layers on top.
The FDE goal
SuitableThe right kind of AI for this job, or no AI where rules do better
SecureClient data protected, access limited, every action traceable
Least costSmallest model and fewest calls that still meet the measure
MaintainableSomeone else can run, fix and change it after you leave
Does the jobMeasured against the outcome it was set up to deliver
What a forward deployed engineer does
An FDE works inside the client's world. They own the path from "how does this work actually happen" to "a running solution the owner trusts". The program trains each step of that loop.
EmbedJoin the client team, meet the process owner and the people doing the work.
DiscoverInterview, shadow, read documents and data; map the real process.
FrameWrite the problem, the success measure and the business case.
DesignLay out two or three options and recommend one.
BuildShip a POC on realistic data with AI coding assistants.
EvaluateTest on real cases; fix what fails; set human review.
Demo and hand overShow the value, agree the next step, leave a runbook.
Six competencies
Discovery: finds the real problem
Business sense: sizes value and cost
AI engineering: prompts, tools, retrieval, agents, evaluation
Run a discovery workshop and produce a process map and brief
Propose solution options with honest trade-offs
Build a working POC with an LLM and their own data
Measure it on a test set and explain its limits
Deploy it safely and demo it to decision makers
Two ways in, one finish line
Track A
Software background
Developers, data engineers, QA and IT staff. You can build; the program teaches you to find the right thing to build and to sell it to the client.
Extra focus: domain and business analysis, discovery, product ownership, communication
Fast-track: Module 3 after a skills check
Track B
Domain or business background
Business analysts, operations, quality, finance and functional consultants. You know the work; the program teaches you to build and ship with AI tools.
Extra focus: code basics, APIs, data, Git, AI-assisted coding, deployment
Fast-track: Module 4 after a skills check
Learners work in mixed pairs from week 1. Each pair has one person from each track, so they teach each other and finish the capstone together.
Prerequisite concepts
Check what you can already explain to a colleague in plain words. Unchecked items show where to start. Your ticks are kept in this browser only.
0 of 42 concepts
Technology
Track B
Track B
Track B
Both tracks
Both tracks
Both tracks
Track A
AI and LLMs
Both tracks
Both tracks
Both tracks
Both tracks
Both tracks
Both tracks
Both tracks
Track A
Development
Track B
Track B
Track B
Both tracks
Track B
Track B
Deployment
Track B
Track B
Both tracks
Both tracks
Both tracks
Product ownership
Track A
Both tracks
Track A
Track A
Track A
Client requirements
Track A
Both tracks
Both tracks
Track A
Both tracks
Track A
Market and business analysis
Track A
Track A
Both tracks
Track A
Track A
The 12-week plan
Modules overlap so the two tracks can work on their extra modules in parallel. Select a bar to open the module.
Both tracks both tracks Track A software-background focus Track B domain-background focus
Training material
Full lessons for every module: a diagram, a plain explanation, a worked example, practice you can start today and links to go deeper. Anyone with basic understanding can follow them alone.
Explain how LLMs work, where they fail, and how to use them well in daily work.
What we cover
Tokens, context windows, temperature, cost
Prompting patterns and structured output
Hallucination, grounding and citations
RAG, tools and agents at a conceptual level
Responsible use: data, privacy, bias, human review
Hands-on lab
Take one messy real document (policy, contract, SOP). Use an LLM to summarize, extract fields into JSON and answer 10 questions. Record where it was wrong and why.
You hand in
A prompt log with 10 tested prompts and a short note on failure patterns.
Read and change code, call APIs, query data and work in Git, with an AI coding assistant beside you.
Software-background learners take a short skills check and skip what they can already do.
What we cover
How the web works
Python basics
Calling REST APIs, JSON, auth
SQL queries and joins
Git, GitHub and pull requests
Using Claude Code or Cursor to explain and change code
Hands-on lab
Write a small script that pulls records from a public API, stores them in SQLite and answers three questions with SQL. Commit it to GitHub through a pull request.
You hand in
A working repository with a README a colleague can run.
From demo to dependable: agentic system design and the FDE portfolio
Both tracks
Turn a working demo into a system a client can trust, and build the portfolio and interview skills that prove you can do it.
What we cover
Where AI belongs: deterministic code, model judgment or human approval
Designing for the unhappy path: schemas, validation, retries, fallbacks
Agentic system design from a vague ask to production
Integrating with existing systems; shadow mode to production
Example portfolio projects and a 30-day build plan
Preparing for FDE interviews: take-homes, live Python, case studies
Hands-on lab
Map one real workflow step by step into code, model and human; harden your POC for its top five failure modes; then run a mock agentic design interview in pairs.
You hand in
A step-by-step operating map, a failure-mode list with how each is handled, and a one-page portfolio write-up of your POC.
Each pair takes one use case for the capstone. Starter cases fit learners new to building; stretch cases add long documents, high stakes or regulation. Mentors provide anonymized or synthetic data for each case.
Core
1. Invoice exception handling
Finance operations
Client
Accounts payable lead at a mid-size distributor. 3,000 supplier invoices a month; about 15% fail matching against purchase orders.
Pain
Clerks chase buyers by email, re-key data and miss early-payment discounts.
Data you get
50 sample invoice PDFs, a purchase order export (CSV), an email thread sample, the AP policy (PDF).
Deliver
Process map of the exception path
Extraction of invoice fields into JSON with a match check
Drafted exception summary and suggested resolution for a clerk to approve
Evaluation on 30+ invoices: extraction accuracy and correct routing
Tests: Document AI, rules plus LLM, human approval, business case
Core
2. Support ticket triage and reply drafting
Customer service
Client
Head of support at a software company. 1,200 tickets a week across email and chat.
Pain
Tickets sit unassigned; agents answer the same questions from scratch.
Data you get
500 anonymized tickets with categories, a help-center export, an SLA policy.
Deliver
Category and priority classifier with confidence
Reply drafts grounded in help articles, with citations
Routing rules and an escalation path
Evaluation: classification accuracy, groundedness of replies
Discovery notes and a process map reviewed by the mock client
One-page brief with the success measure and a sized business case
Options paper with a recommendation
Working POC in a repository with a README
Evaluation set of 30 or more cases with results and the top failures explained
Deployed URL behind access control
Ten-minute demo and a hand-over pack: runbook, owner, next step
Rubric and graduation
Dimension
Not yet
Practitioner
FDE-ready
Discovery
Asked mostly about features or opinions; process map has gaps.
Captured the real process and pain from the last real cases; map is complete.
Found the root problem and the hidden constraint; the client recognizes the map as their work.
Problem framing and value
Success is a feature list.
Success is a measurable outcome with a baseline.
Business case is sized, credible and agreed by the process owner.
Solution options
One option, no trade-offs.
Two or three options with clear pros, cons and costs.
Recommendation is defensible on accuracy, cost, risk and adoption; knows when not to use AI.
Build quality
Works only on the demo case.
Works on realistic data; code is readable and runs from the README.
Clean, small, testable; another person can extend it in a day.
Evaluation and safety
No test set.
30+ cases, a pass rate and the top failures explained.
Evaluation drives changes; human review and access limits match the risk.
Deployment and operations
Runs only on a laptop.
Deployed behind access control with secrets handled properly.
Tracing, cost tracking and a runbook; someone else could operate it.
Communication
Demo is a tour of features.
Demo starts from the client problem and shows real results.
Handles hard questions calmly; states limits honestly; next step is agreed.
To graduate: FDE-ready in Discovery, Build quality, Evaluation and safety, and Communication, and at least Practitioner in the rest. After graduation, a new FDE shadows a live engagement, then co-leads one, then leads.
Certifications
Optional, and never a substitute for the capstone. Pick one or two that match your track and the platforms your clients use.