How I Applied to the FDE Position at Anthropic

I recently applied for Anthropic’s Forward Deployed Engineer, Applied AI position. The role is posted in the Greater Syracuse area — right where I live. It felt like the kind of opportunity you don’t sit on, so I didn’t.

I used Claude Code to build the entire application. This post is a transparent walkthrough of how I did it, what worked, and what the process actually looked like. Following Anthropic’s candidate AI guidance, I drafted all content myself and used Claude for refinement, architecture decisions, and implementation. Nothing is fabricated.

What I built

The application isn’t just a resume and a cover letter. I built a full landing page at jessepeplinski.com/anthropic with:

• An intro video where I walk through my background and why I’m applying
• A “Why Anthropic” section explaining what drew me to the role and company
• A requirements mapping showing how my experience maps to every line in the job posting
• Systems thinking examples from my work at Equitable, Hack Upstate, and side projects
• Community impact metrics from running Careers in Code (74 graduates, approximately $1.1M in grant-backed program funding managed)
• A transparency section about using AI in the application process

I also rebuilt my personal website from scratch (jessepeplinski.com), wrote a blog post on modern software engineering, and created a print-friendly resume page that I could Cmd+P into a PDF — all in the same push. No Google Docs, no Word templates. The resume lives as an Astro page with its own print-optimized CSS, branded with the same orange accent as the rest of the site.

Setting up the context

Before I started building anything, I set up a project folder with all the source material Claude would need. The quality of the output depends entirely on the quality of the input — so I was intentional about this.

The folder structure looked like this:

FDE/
  content/              # Markdown source files (resume, CV, LinkedIn)
  jesse-performance-reviews/  # 8 years of Equitable performance reviews (2017-2025)
  peptech.dev/          # Existing business/portfolio website
  jessepeplinski.github.io/   # Personal website (to be rebuilt)
  CLAUDE.md             # Project instructions for Claude Code

The CLAUDE.md file was key. It told Claude what I was applying for, where the source materials lived, that I needed to follow Anthropic’s candidate AI guidance, my official job title, and that the content should match my writing voice. I also included 8 years of performance reviews from Equitable — these are my own reviews about my performance, not proprietary data. No customer information, no company IP, no PII. They were the primary source material for surfacing career growth that wasn’t visible on my outdated resume.

The content pipeline was simple: resume.md was the foundation, which fed into the LinkedIn profile, the website, and the /anthropic page. Everything traced back to real source material.

How Claude Code was involved

I used Claude Code for nearly every part of the technical implementation. Here’s what that looked like in practice:

Site architecture: Claude helped me scaffold the Astro project, set up Tailwind 4, configure the build pipeline, and deploy to Netlify. I described what I wanted and we iterated together on the structure.

Content refinement: I wrote all the raw content — my resume bullets, the “Why Anthropic” narrative, the blog posts. Claude helped me tighten the writing, catch inconsistencies, and organize sections.

Component development: The site uses reusable Astro components (HighlightCard, TechBadge, Header, Footer). Claude built these based on the design system I described — dark mode, orange gradient brand, Inter font.

Code I didn’t write by hand: Basically all of it. The HTML, CSS, Astro templates, Tailwind config, build configuration — Claude wrote the code. I directed the architecture, made design decisions, reviewed the output, and course-corrected when things didn’t feel right.

What I actually did vs. what Claude did

To be direct about it:

Me:

• All the raw content and ideas — every claim, every metric, every story
• Design direction and aesthetic decisions
• Voice and tone (Claude adapted to mine, not the other way around)
• Final review and approval of everything
• Recording the video
• The decision of what to include and what to cut

Claude:

• All code implementation
• Content refinement and organization
• Catching factual inconsistencies
• Building the site infrastructure

What I learned

AI doesn’t replace the thinking — it accelerates the execution. Every idea, every claim, every design decision came from me. Claude made it possible to turn all of that into a shipped product in a fraction of the time it would have normally taken.

The skill that matters is communication. The better I could articulate what I wanted, the better the output. When I was specific, things came together fast.

Ship imperfect. I could have spent another week polishing. Instead, I shipped it and moved on. The page is live, the video is up, and the application is in. Done is better than perfect.

Why I’m sharing this

Anthropic’s candidate AI guidance says to be transparent about how you use AI. I’m taking that seriously. This isn’t a disclaimer — it’s a demonstration. This is exactly what a Forward Deployed Engineer does: uses AI tools effectively to ship real things, understands their strengths and limitations, and is honest about the process.