Build it first. Then take it to production.

Become a
Forward Deployed Engineer

Twelve weeks, live with industry experts, turning one client's business problem into a working AI system you deploy, measure and hand over - the project you walk a hiring manager through.

The stack you build with

  • Python
  • FastAPI
  • React
  • TypeScript
  • OpenAI
  • Claude
  • LangChain / LangGraph
  • Postgres + pgvector
  • Docker
  • GitHub Actions
  • AWS
  • Azure
12weeks, live
18modules
45hands-on labs
17checkpoint deliverables
1capstone and Demo Day

Part engineer, part architect, part consultant

A Forward Deployed Engineer sits with the client and owns the whole arc, from the first interview to the handover. This program follows that same arc, week by week.

  1. 01DiscoverThe problem, the users, the data
  2. 02DesignScope and architecture
  3. 03BuildThe full-stack agentic app
  4. 04DeployContainers, CI/CD, their cloud
  5. 05MeasureEvals, monitoring, cost
  6. 06Hand overDocs, training, adoption

Hired by OpenAI, Anthropic, Palantir, Salesforce, Databricks and Deloitte - and every enterprise taking AI to production.

What Forward Deployed Engineers are paid

One of the fastest-growing titles in AI, and the pay reflects how few people can do the job.

  • Forward Deployed Infrastructure EngineerArtmac Soft$201,513
  • Forward Deployed Engineer - DatabricksDeloitte$192,480
  • AI Engineer, Forward Deployed EngineeringArtmac Soft$186,336
  • Forward Deployed Data EngineerArtmac Soft$173,717
  • Forward Deployed Engineer - PalantirDeloitte$149,969
$150K–$250K+typical range for the role
AI + cloud + clientthe three skill sets every post asks for
~25% travelworking on-site with customers

Posting figures are estimated annual salary, USD, Houston TX, via Adzuna. The range reflects the role across markets and seniority.

THE CURRICULUM

See exactly what you’ll build, week by week

Get a clear look at the learning path, hands-on work, and the real-world projects that take you from AI foundations to deploying production-ready solutions as a Forward Deployed Engineer.

12-week live program Hands-on labs Real-world capstone

18 modules across 12 weeks

Two hours a day, Monday to Friday. The first hour is concept and a live demo; the second is a lab you build yourself. Open any module to see what you cover, the labs inside it, and what you hand in.

What you cover

  • Python foundations for AI engineering
  • Typing, packaging and project structure
  • Tests you can put in CI

Labs

  • Build a typed Python utility for an AI service
  • Write the pytest suite that covers it
  • Package the project and run its tests in CI
CheckpointPython utility + tests

What you cover

  • The delivery lifecycle for AI projects
  • AI-assisted coding workflows
  • Git, branches and pull requests

Labs

  • Ship a change through a branch, pull request and review
  • Drive an AI coding assistant through a real refactor

What you cover

  • FastAPI services and OpenAPI contracts
  • SQL and PostgreSQL fundamentals
  • Data modeling for AI applications

Labs

  • Build the Ticket Service API with FastAPI
  • Model and query the data in PostgreSQL
  • Publish the OpenAPI contract and test every endpoint
CheckpointTicket Service API

What you cover

  • OpenAI and Anthropic Claude APIs
  • Prompt engineering and structured output
  • Bedrock, Azure OpenAI and Ollama options

Labs

  • Call the OpenAI and Claude APIs side by side
  • Force structured output your app can rely on
  • Build a provider-agnostic LLM layer and swap models
  • Compare cost and latency across providers
CheckpointLLM service + chat UI

What you cover

  • React, Vite and TypeScript
  • A streaming chat interface
  • An approval queue for human-in-the-loop

Labs

  • Build the streaming chat interface in React
  • Add the human approval queue
  • Wire the UI to your FastAPI service

What you cover

  • Chunking, embedding and the pgvector store
  • Hybrid search and reranking
  • Citations, streaming and groundedness scoring

Labs

  • Chunk and embed a document set into pgvector
  • Add hybrid search across keyword and vector
  • Add cross-encoder reranking
  • Return streamed answers with citations
  • Score the pipeline with RAGAS groundedness
CheckpointKnowledge assistant (RAG)

What you cover

  • LangChain chains and runnables
  • Retrieval orchestration
  • Connecting tools and data sources

Labs

  • Build retrieval chains with LCEL
  • Connect tools and data sources to a chain
  • Compare chain designs on the same question set

What you cover

  • LangGraph state, nodes and conditional edges
  • Triage → plan → act agent loops
  • Human-approval interrupts, memory and traces

Labs

  • Build a StateGraph with typed state
  • Add conditional edges and routing
  • Parallelise work with fan-out and fan-in
  • Add a human-approval interrupt
  • Persist memory across turns
  • Trace and visualise the graph
CheckpointMulti-agent workflow

What you cover

  • OpenAI Agents SDK and Claude SDK
  • Model Context Protocol servers and clients
  • Tool calls into Jira and Confluence

Labs

  • Build your first ReAct agent with tools
  • Write a custom MCP server for Jira
  • Call Confluence through MCP from the agent
  • Compare the OpenAI Agents SDK with the Claude SDK
CheckpointMCP server

What you cover

  • LangSmith tracing, RAGAS and DeepEval
  • Guardrails and Presidio PII redaction
  • OWASP LLM Top 10 and red-teaming

Labs

  • Build a golden dataset from real tickets
  • Score groundedness and relevance with RAGAS
  • Check tool-call accuracy with DeepEval
  • Redact PII with Presidio
  • Red-team for prompt injection and data leakage
  • Trace every run in LangSmith
CheckpointEval + red-team report

What you cover

  • Docker images for API, agent, UI and Postgres
  • GitHub Actions pipelines
  • Eval gates, lint, unit and integration tests

Labs

  • Containerise the API, agent, UI and Postgres
  • Build a GitHub Actions pipeline
  • Add the eval gate that blocks a bad deploy

What you cover

  • Render, AWS ECS / Fargate and Azure Container Apps
  • Private networks, managed secrets and SSO
  • Monitoring, logs, traces and cost

Labs

  • Deploy to AWS ECS / Fargate or Azure Container Apps
  • Set up private networking, secrets and SSO
  • Build the Grafana KPI dashboard
CheckpointCI/CD + cloud deploy

Every module is a layer of the same application

Nothing is taught in isolation. By week 12 these layers are one running system in a cloud account.

Front endM5
React · Vite · TypeScript
API layerM1 · M3
FastAPI · Python · OpenAPI
Agent layerM7–M9
LangGraph · LangChain · Agents SDK · Claude SDK · MCP
LLM providersM4
OpenAI · Claude · Bedrock · Azure OpenAI · Ollama
Data & retrievalM3 · M6
PostgreSQL · pgvector · Redis · hybrid search
Quality & securityM10
LangSmith · RAGAS · DeepEval · Presidio · Pytest
Ship & runM11 · M12
Docker · GitHub Actions · Render · AWS · Azure · Grafana

Pick a client problem, or bring your own

Your capstone runs through all six steps and ends at Demo Day, in front of a panel.

🎧

IT Helpdesk Copilot

Knowledge-base answers plus ticket actions

📄

Claims Triage Agent

Document extraction with human approval

👤

HR Onboarding Assistant

Policy answers and joiner checklists

🧾

Invoice Reconciliation

Match invoices to POs through an ERP API

📊

Sales Account Intelligence

CRM research and meeting briefs

🧩

Bring your own

A problem from your own work, instructor-approved

What you walk away with

Not a certificate - a deployed system and the paperwork around it. Here is what one capstone produced.

38%tickets deflectedagreed target was 35%
41sfirst responsedown from 6 hours
0.93groundednessscored on 120 real tickets
$0.012per conversationbudget was $0.05
🔗

Live URL and GitHub repo

The deployed app, its CI/CD history, tests and eval suite

📐

Architecture document

C4 diagrams and ADRs explaining every design decision

📈

Eval report

Groundedness, relevance, tool accuracy, red-team findings and fixes

📋

Client artifacts

The problem brief, PRD-lite and SOW outline

📦

Handover pack

Runbook, user guide and adoption plan for the client team

🎥

Demo Day recording

A 10-minute executive presentation you can send a recruiter

Who this is for

Working professionals who can give it two hours a day. No prior AI or ML experience needed.

  • 💻
    Software developersBackend and full-stack engineers moving into applied AI
  • 🧭
    Consultants & tech leadsClient-facing delivery and solution roles
  • 📊
    Data & ML engineersYou know models and pipelines; now you need production skills
  • 🏢
    Enterprise app specialistsERP, HCM and CRM people building AI extensions

What you should already know

  • One programming language, ideally Python
  • The command line and basic Git
  • A basic understanding of APIs and databases
  • Three to four hours of self-practice a week

What you need to have

  • A laptop with 8 GB RAM, 16 GB is better
  • Admin rights to install Docker and Node.js
  • OpenAI and Anthropic API accounts
  • GitHub, Render and an AWS or Azure account

Six reasons this program is different

🔨

Build first

More than half the time is labs and build sprints, not slides.

🔀

Two model providers

OpenAI and Claude, swapped freely - no vendor lock-in.

🧱

The whole stack

React, FastAPI and pgvector through to a live URL.

🛡️

Production ready

Evals, guardrails, observability and the OWASP LLM Top 10.

🤝

Client facing

Discovery, demos and handover practised, not just discussed.

📂

Proof of skill

17 deliverables and one deployed capstone you can walk through.

Where this takes you

Four roles the same skill set opens. Outcomes depend on your effort, your prior experience and the market.

🚀

Forward Deployed Engineer

Embed with clients and ship AI into their environment

Salesforce · OpenAI · Anthropic · Palantir
⚙️

Applied AI Engineer

LLM, RAG and agent features inside products

Product and platform teams
🎯

AI Solutions Engineer

Design and demo AI solutions for customers

Pre-sales and solution teams
🤖

Agentic AI Developer

Multi-agent workflows on enterprise tools

Consultancies and enterprise IT

From where you are today to an offer

The three steps our 46,000+ learners have used, mapped to what the job description asks for.

01

Skills

Not confident, then confident

  • 18 modules and 45 hands-on labs
  • 17 checkpoint deliverables
  • A capstone with a live URL and repo
Both plans
02

Visibility

No calls, then interview calls

  • A resume written in FDE keywords
  • LinkedIn profile rebuilt and optimised
  • A targeted search with weekly alerts
Job preparation
03

Offer

Interviews, then an offer

  • Unlimited mock interviews, technical and behavioural
  • Walking a panel through your own system
  • Negotiation, then a year of on-job support
Job preparation

One program. Two plans.

Both include the full 12-week live program, 45 labs, 17 checkpoints and the capstone.

Build the system
Upskill - program only
$2,997 $1,997

One payment. Learn the stack and ship the system.

  • 12-week live program, led by industry experts
  • 18 modules and 45 hands-on labs
  • 17 checkpoint deliverables
  • Capstone project and Demo Day
  • 1-year access to recordings and labs
  • Community access for a year
Enroll in Upskill - $2,997 $1,997

Six-month money-back guarantee

Flexible
Upskill - program only
$1,097 × 3 $797 × 3

Three monthly payments, $2,097 in total.

  • 12-week live program, led by industry experts
  • 18 modules and 45 hands-on labs
  • 17 checkpoint deliverables
  • Capstone project and Demo Day
  • 1-year access to recordings and labs
  • Community access for a year
Enroll - $1,097 × 3 $797 × 3

Six-month money-back guarantee

Most popular
Upskill + job preparation
$3,997 $2,997

One payment. Learn, ship, and go after the job.

  • 12-week live program, led by industry experts
  • 18 modules and 45 hands-on labs
  • 17 checkpoint deliverables
  • Capstone project and Demo Day
  • 1-year access to recordings and labs
  • Community access for a year
  • Everything above, plus
  • FDE resume preparation and a weekly resume clinic
  • LinkedIn profile rebuilt and optimised
  • Unlimited mock interviews, technical and behavioural
  • Job-search strategy and application tracking
  • 1-year on-job support after placement
Enroll with job preparation - $3,997 $2,997

Six-month money-back guarantee

Most popular
Upskill + job preparation
$1,097 × 4 $797 × 4

Three monthly payments, $3,147 in total.

  • 12-week live program, led by industry experts
  • 18 modules and 45 hands-on labs
  • 17 checkpoint deliverables
  • Capstone project and Demo Day
  • 1-year access to recordings and labs
  • Community access for a year
  • Everything above, plus
  • FDE resume preparation and a weekly resume clinic
  • LinkedIn profile rebuilt and optimised
  • Unlimited mock interviews, technical and behavioural
  • Job-search strategy and application tracking
  • 1-year on-job support after placement
Enroll - $1,097 × 4 $797 × 4

Six-month money-back guarantee

Six months. Love it or leave it.

Do the work, and if it does not deliver, you get your money back.

  • Attend the live sessions and complete the labs
  • Submit all 17 checkpoint deliverables
  • Build and present the capstone at Demo Day
  • Ask for support when you need it - do not go quiet

Did all that and still not satisfied? A full refund. Action-based, six months, no fine print.

Before you enroll

Software developers, consultants and tech leads, data and ML engineers, and enterprise application specialists - anyone who can commit two hours a day, five days a week. No prior AI or ML experience is required; the program teaches LLMs, RAG and agents from the ground up.

One programming language (Python is ideal), the command line and basic Git, and a basic understanding of APIs and databases. On the hardware side: a laptop with 8 GB RAM (16 GB is better), admin rights to install Docker and Node.js, OpenAI and Anthropic API accounts, and GitHub, Render and an AWS or Azure account. API and cloud usage is pay-as-you-go, and free tiers and credits apply where available.

Two hours a day, Monday to Friday, across 12 weeks. The first hour is concept and a live demo; the second is a guided lab you build yourself. Friday is checkpoint review. Plan another three to four hours of self-practice a week.

Live and instructor-led with industry experts. Every session is recorded, and you keep access to the recordings and labs for a year.

A deployed capstone and the artifact set around it: a live URL and GitHub repo, an architecture document with C4 diagrams and ADRs, an eval report, the client artifacts, a handover pack, and a 10-minute Demo Day recording. Every item is something a hiring manager can open.

Both include the full 12-week live program, 45 labs, 17 checkpoint deliverables and the capstone. Upskill + job preparation adds FDE resume preparation and a weekly resume clinic, LinkedIn optimization, unlimited mock interviews, a job-search strategy with application tracking, and a year of on-job support after placement.

Yes. Both plans can be split over three monthly payments - switch to Pay monthly in the pricing section to see the amounts.

No one honestly can. What we guarantee is the work: the program, the capstone, and in the job preparation plan the resume, interview and job-search support behind it. Outcomes depend on your effort, your prior experience and market conditions. The program is also covered by a six-month money-back guarantee.

Yes. Corporate cohorts run at a time slot agreed with you, with capstones set in your own business context, weighted deliverables to track progress, and a readiness checklist before week one. Contact your program advisor to arrange it.

Take AI from idea to production

Join the next cohort, build a real AI system, and walk away with a project a hiring manager can open.

Forward Deployed Engineer Program · K21 Academy