What a forward deployed engineer does
The job
You work directly with a customer. You scope a messy real-world problem, build the AI system that solves it, and ship it into production.
Why now
AI companies, and the enterprises adopting AI, are hiring engineers who can do this.
Why it lasts
The job combines engineering, judgment and client communication. That mix is harder to commoditize than writing code alone.
Why a residency
The role is learned by doing it, under real deadlines and alongside other people. Online courses alone rarely give you that.
Who should apply
Recent engineering graduates
Computer science and engineering graduates looking for a first role
You have the degree. Now get the offer. Turn coursework into shipped, deployed work, and graduate with a portfolio of client-style projects that gets interviews.
Experienced IT professionals
Enterprise, infrastructure and consulting professionals, including those reskilling after a layoff
Your experience is an asset. Make it AI-ready. Client-facing and systems experience is an advantage in this role, and the residency turns it into the FDE profile.
Who it is not for
People who cannot commit full time and in person for all 12 weeks. This is a full-time commitment, like a job.
Four weeks of foundations, then a project every week
Four weeks of foundations, taken from the curriculum of our LLM and agentic AI bootcamps, then four build sprints with code reviews and demos, a client capstone, and demo day. It is a full-time commitment, like a job: in-class learning, projects, homework and talks from industry experts.
Foundations
Week 01Mon, Jan 18How LLMs workTokens, embeddings and attention
How large language models work
The map of the stack you will deploy for clients, then what happens inside the model. Tokenization, embeddings and how models generate text; transformers from the ground up, with positional encoding and self-attention; where models are strong and why they hallucinate; and how pre-training, fine-tuning and in-context learning each give a model your own knowledge.
Week 02Mon, Jan 25Retrieval and contextVector search, agentic RAG and fine-tuning
Retrieval, context and fine-tuning
Retrieval sets the ceiling on everything a client sees. Embeddings and vector databases; semantic, keyword and hybrid search; HNSW indexes and re-ranking; agentic RAG with multi-step retrieval; the problems production adds, from PII-safe ingestion to latency and cost; and when fine-tuning with LoRA beats better retrieval, and when it cannot help.
Week 03Mon, Feb 1Building agentsLangChain, LangGraph, design patterns and MCP
Building agents
From chains to agents that hold state. LangChain components, tools and LCEL; LangGraph graphs, typed state and memory; the reflection, planning, tool-use and multi-agent patterns, and what each costs in control; and Model Context Protocol, so a tool is exposed once and called by any agent, with A2A and ACP compared.
Week 04Mon, Feb 8Evaluation and productionEvals, observability, guardrails and deployment
Evaluation, observability and deployment
Prove a system works, then keep it working. Benchmarks, RAGAs and G-Eval; traces and online evaluation once it is live; the reference architectures that recur in production, from a single agent with tools to supervisor and workers and human-in-the-loop review; agents in Docker sandboxes with policy enforcement and guardrails; and deploying to the cloud.
Build sprints
Week 05Mon, Feb 15Sprint 1A governed document-review agent
Build sprint 1: a governed document-review agent
Your first client-style engagement: a document-review agent for a public-sector client, with the governance that client needs. Scope it, build it and ship it inside the week, then demo it and take a code review.
Week 06Mon, Feb 22Sprint 2A customer support assistant
Build sprint 2: a customer support assistant
A customer support assistant with evals, guardrails and monitoring. Ship it inside the week, demo it, and take a code review.
Week 07Mon, Mar 1Sprint 3An agent inside an enterprise system
Build sprint 3: an agent inside an enterprise system
A data pipeline and an agent workflow, integrated into an existing enterprise system rather than built on a clean slate. Ship it inside the week, demo it, and take a code review.
Week 08Mon, Mar 8Sprint 4A human-in-the-loop operations agent
Build sprint 4: a human-in-the-loop operations agent
A multi-agent workflow that works incoming requests, such as invoices, claims or orders, across several systems over MCP, and routes the decisions that matter to a person for approval. Ship it inside the week, demo it, and take a code review.
Client capstone
Week 09Mon, Mar 15Capstone: scopeStakeholders, constraints and a plan
Client capstone: scope the problem
The capstone is a real problem with stakeholders and constraints. Week one is the part of the job courses skip: meet the stakeholders, pin down what they actually need, and agree on what you will ship.
Week 10Mon, Mar 22Capstone: buildBuild it and evaluate it
Client capstone: build and evaluate
Build the system you scoped, measure it against what the stakeholders agreed to, and keep them informed as the constraints push back.
Week 11Mon, Mar 29Capstone: deployInto production
Client capstone: deploy to production
Ship the capstone into production, with the evaluation, security and governance the client signs off on.
Demo day and hiring
Week 12Mon, Apr 5Demo dayDemo day and hiring-partner interviews
Demo day and hiring
Present your capstone at demo day, then interview with hiring partners, with mock interviews and support through your offers.
What you'll build
A portfolio of client engagements, not tutorials
Every project is framed as work for a client, with a brief, constraints and someone to answer to. You finish with four shipped sprint projects and a capstone in production, and you can explain the decisions behind each one.
- Governed document-review agent
- Customer support assistant
- Agent workflow in an enterprise system
- Human-in-the-loop operations agent
- Your client capstone
What each project proves
- Scoping
- Turn a messy brief into a plan
- Stakeholders
- Keep a client informed and aligned
- Deployment
- Ship to production, not a notebook
- Evaluation
- Measure it before the client asks
- Governance
- Security and governance built in
What happens in week 12 and after
This is the first cohort, so there are no placement numbers yet. When there are, we will publish them with the method behind them: cohort size, how many graduates were job-seeking, the timeframe, and the roles counted.
Demo day
Present your capstone to hiring partners and the cohort.
Partner interviews
Interview with hiring partners in week 12.
Interview and offer support
Mock interviews before, and help negotiating your offer after.
Offers start at $125,000
Every offer from a hiring partner starts at $125,000 a year or more. It is a floor, not a promise of an offer.
Free to cohort members. Hiring partners pay.
The program is funded by the companies that hire its graduates, not by you.
Free to cohort members
What you pay
You pay nothing to attend, and cover only your own travel, housing and living costs. No fees, clawbacks or repayments, whether you finish, leave or turn down an offer.
Hiring partners fund it
How it works
Companies that need forward deployed engineers fund the program. They pre-pay for hiring credits, and we are paid only when they hire and keep one of our graduates.
Why it works for you
Aligned incentives
We only succeed when you get hired, and the companies you interview with in week 12 are the ones funding your seat.
Three months in Redmond/Seattle
The program is free, but you cover your own travel, housing and living costs in the Redmond/Seattle area, so budget for three months.
Housing
We connect admitted students who want to share housing in the Redmond/Seattle area.
Getting there
Classes are at 5010 148th Ave NE, Suite 200A, Redmond, WA 98052, in the Seattle area.
Work authorization
Open to applicants authorized to work in the US. Sponsorship for later cohorts depends on our hiring partners.
A selective, competitive admissions process
Admission is highly selective and competitive: 30 seats a cohort, each filled through a technical assessment and an interview. We look for problem-solving, communication, grit and commitment.
1. Apply online
Tell us about your background and why you want this role. Applications close December 21, 2026, and shortlisted applicants hear from us within two weeks.
2. Technical assessment
A take-home task that looks like the job: from a short client brief, build a small working AI application, then write a one-page note to the client explaining your choices and what you would do next.
3. Interview
A technical and motivation interview.
4. Offer and enrollment
Offers follow the interview, and the cohort starts January 18, 2027.
Apply for the January 18, 2027 cohort
Applications close December 21, 2026. Shortlisted applicants hear from us within two weeks, for a technical assessment and an interview.
The full link, starting with https://
Or a link to other work you have shipped.
Your background, something you have built, and why you want to be a forward deployed engineer.
We have your application. If you are shortlisted, our admissions team will reach out within two weeks about the technical assessment.
4.95 from 660+ verified reviews
The most thrilling revelation for me was the multi-step problem-solving approach that mirrors human thinking, offering limitless applications with creative usage of LLMs. I now feel empowered to tackle complex challenges at work with the knowledge gained from this enriching experience.
Aishwarya Raman — read the full review (opens in a new tab)Senior Software Engineer, Microsoft AzureThe Data Science Dojo LLM Bootcamp is a hands-on experience, delving into LLMs and their real-world applications. Expert instructors provide a comprehensive framework with a practical focus, enabling you to fine-tune, evaluate, deploy, and monitor LLM applications.
Bernard Tumanjong — read the full review (opens in a new tab)Chief Operations Analysis, US Space Force Coming from a linguistics background, the Large Language Model Bootcamp provided me with cutting-edge insights into language and data science. Now equipped with practical knowledge, I'm excited to implement these learnings at Starbucks. Whether you're well-versed in development or just starting with Python, this boot camp caters to diverse skill levels.
Erika Davis — read the full review (opens in a new tab)Senior Business Reporting Analyst, StarbucksExceptional, hands-on initiation into LLMs and practical applications.
Jeff Fong — read the full review (opens in a new tab)Principal Product Manager, HOVER IncSeamless navigation and invaluable hands-on exercises.
Maryam Bagheri — read the full review (opens in a new tab)Senior Data Scientist, Deloitte
Frequently asked questions
How much does the FDE bootcamp cost?
Why is the program free?
How competitive is admission?
What will offers pay?
What happens after I apply?
When does the next cohort start?
Can I work part-time during the program?
Is there a remote or hybrid option?
What prior experience do I need?
Is there pre-work before the residency starts?
What does a typical day look like?
Do I owe anything if I turn down an offer?
What happens if I don't get an offer?
What if I need to leave the program early?
Do I need my own laptop? Are cloud credits provided?
Can international applicants join?
I hire engineers. How do I become a hiring partner?
The next cohort starts January 18, 2027
Highly selective: 30 seats, free to cohort members. Applications close December 21, 2026.