Forward Deployed Engineer (FDE) Roadmap

A curated public repository for learning, building, and mastering the skills required to become a Forward Deployed Engineer (FDE) at frontier AI companies.

2026 FDE Roadmap

---

What is a Forward Deployed Engineer?

A Forward Deployed Engineer (FDE) is a hybrid role combining:

• Software Engineering
• AI Systems Engineering
• Product Thinking
• Technical Consulting
• Customer Deployment

FDEs work directly with organizations to:

• Understand real business problems
• Design AI-powered solutions
• Deploy systems in production environments
• Build agentic workflows around LLMs
• Ensure observability, safety, and evaluation

This repo documents the complete learning journey toward becoming an industry-ready FDE.

---

Repository Goals

This repository aims to become:

• A complete FDE learning reference
• A practical implementation hub
• A portfolio-building guide
• A deployment-pattern library
• A collection of production-grade AI engineering examples

---

Core Learning Pillars

1. LLM-Native Foundations

Learn how modern AI systems are engineered around stochastic LLM behavior.

Topics

• Workflows vs Agents
• Context Engineering
• Structured Outputs
• Tool Use
• Prompt Caching
• Reasoning Models
• Context Windows
• Streaming APIs

Resources

• Anthropic Engineering Blogs
• OpenAI Agents Guides
• Chip Huyen — AI Engineering
• Simon Willison Blog
• Hamel Husain Articles

---

2. Major API Surfaces

Become fluent with frontier AI APIs.

APIs to Learn

• OpenAI
• Anthropic Claude
• Google Gemini
• Meta Llama
• Open-weight models

Key Concepts

• Chat/message loops
• Function calling
• Tool usage
• JSON mode
• Batch APIs
• Extended reasoning
• Cost optimization

---

3. Prompt & Context Engineering

Master production-grade prompting patterns.

Topics

• Few-shot prompting
• Chain-of-thought
• ReAct
• Reflexion
• XML tagging
• Self-consistency
• Plan-and-execute
• Structured prompting

---

4. Agent Infrastructure & Protocols

The backbone of enterprise AI systems.

Topics

• MCP (Model Context Protocol)
• Agent Skills
• Multi-agent systems
• Orchestrators
• Sub-agents
• Agent memory
• Tool routing

Frameworks

• OpenAI Agents SDK
• LangGraph
• CrewAI
• AutoGen
• Mastra

---

5. Retrieval & RAG

Understand retrieval systems deeply.

Topics

• Chunking strategies
• Hybrid search
• Dense retrieval
• BM25
• Reranking
• Recall@K
• MRR
• Graph-based retrieval

Tools

• Pinecone
• LlamaIndex
• Vector Databases

---

6. Evaluation Frameworks

One of the most important production skills.

Topics

• LLM-as-a-judge
• Golden datasets
• Regression testing
• Pairwise comparison
• Rubric-based grading
• Online evaluation systems

Tools

• Promptfoo
• Braintrust
• Langfuse
• Inspect AI
• OpenAI Evals

---

7. Observability, Security & Safety

Production AI systems require reliability.

Observability

• Tracing
• Token logging
• Cost monitoring
• Latency tracking
• Prompt tracking

Security

• Prompt injection prevention
• OWASP LLM Top 10
• AI Risk Management
• Secure deployment patterns

Tools

• Langfuse
• Helicone
• Arize Phoenix
• Braintrust

---

8. Vertical Specialization

Choose one industry and go deep.

Suggested Verticals

• Finance
• Healthcare
• Legal
• Customer Experience (CX)

Example Concepts

Finance

• KYC/AML
• Trade lifecycle
• SR 11-7

Healthcare

• HIPAA
• PHI
• FDA SaMD

Legal

• Contract review
• Document retention
• Legal privilege

CX

• Ticketing systems
• Agent assist
• Voice AI
• Deflection systems

---

9. The Consulting Motion

FDEs are not just engineers.

Skills

• Discovery calls
• Stakeholder communication
• Executive presentations
• Requirement scoping
• Writing case studies
• Handling ambiguity

Recommended Reading

• The Mom Test
• SPIN Selling
• The Pyramid Principle
• High Output Management
• An Elegant Puzzle

---

10. Evidence Portfolio

Your portfolio matters more than certificates.

Recommended Projects

• MCP Server
• Agent Skill
• Multi-agent workflow
• Eval harness
• Deployment architecture writeups
• Enterprise AI demos
• AI workflow automation systems

---

---

---

Suggested Repository Structure

fde-roadmap/
│
├── README.md
├── resources/
├── notes/
├── projects/
│   ├── mcp-server/
│   ├── agent-skills/
│   ├── eval-harness/
│   ├── multi-agent/
│   └── vertical-demos/
│
├── case-studies/
├── architectures/
├── deployment-patterns/
└── demos/

---

What NOT to Over-Invest In

Avoid Spending Too Much Time On:

Building Vector DBs From Scratch

Focus on deploying AI systems effectively.

Fine-Tuning Everything

Most production systems rely more on prompting and context engineering.

Heavy Framework Dependency

Understand frameworks, but prioritize direct API understanding.

Chasing Every New Model

Focus on capability classes instead of hype cycles.

---

Recommended Learning Resources

Core Resources

• Anthropic Engineering
• OpenAI Docs
• Gemini API Docs
• Simon Willison Blog
• Hamel Husain
• Eugene Yan
• Pinecone Learning Center
• Inspect AI

---

Future Plans for This Repository

• [ ] MCP Server Implementations
• [ ] Agent Skill Library
• [ ] RAG Experiments
• [ ] Evaluation Benchmarks
• [ ] Multi-Agent Architectures
• [ ] Production Deployment Examples
• [ ] Security & Safety Guides
• [ ] Case Studies
• [ ] Architecture Diagrams
• [ ] Interview Preparation

---

Contributing

Contributions, improvements, and discussions are welcome.

If you’re also learning AI Engineering, Agent Systems, or the FDE pathway, feel free to contribute.

---

Disclaimer

This repository is a community-driven educational resource inspired by the public FDE roadmap shared by multiple resources.

All credit for the roadmap concept belongs to its creator.

---

License

MIT License