How to pick an AI job, and land it.

Claude Code is Claude running in your Terminal. It can read files on your computer, write new ones, run commands, and (in step 02) drive your Chrome browser. That means it can do the research, the spreadsheets and the typing of a job search while you make the decisions.

1. Open Terminal: press ⌘ Space, type "Terminal", press Return.
2. Paste the official installer and press Return:

   curl -fsSL https://claude.ai/install.sh | bash

   copy
3. Make a folder for your job hunt and start Claude inside it:

   mkdir ~/ai-job-hunt
   cd ~/ai-job-hunt
   claude

   copy
4. The first time, a browser window opens. Log in with your Claude Pro or Max account.
5. Drag your résumé PDF into the ai-job-hunt folder in Finder (it's in your home folder).

1. In Chrome, open the Chrome Web Store, search Claude, and install the official extension from Anthropic. Sign in with the same Claude account.
2. In that same Chrome, log in to LinkedIn yourself. Claude works inside your logged-in session. It never needs your password.
3. Back in Terminal, start Claude with browser access:

   claude --chrome      # or type /chrome inside a running session

   copy
4. Run a test:

   Open a new Chrome tab, go to my LinkedIn profile, and tell me what a recruiter sees in the first 5 seconds. Be blunt.

You'll watch Claude open tabs, read pages, click and type. Chrome asks you to approve each website the first time. You can stop it at any moment.

"AI jobs" is not one job. Recruiters search for a specific title, so your profile has to be built around one target role (plus one backup). Start from what you can already prove.

How to choose: the 3-circle test

1. Proof today. Which role can you back up with real work, right now or within 2 weeks?
2. Demand. Which role has the most open posts in your city or remote? (Step 04 measures this.)
3. Staying power. Which work would you happily do for 2+ years?

Your target is where all three overlap. Career changers usually land fastest in a role that reuses their old industry: a nurse in AI evaluation for healthcare models, a salesperson as an AI solutions engineer, an ops manager in AI automation.

Recruiters search LinkedIn by job title, skills, location, years of experience, and whether you're Open to Work, then add keywords. They write those keywords by copying from the job post the hiring manager gave them. So the job posts tell you exactly what they'll type.

Collect the posts

Turn them into a keyword list

How to read the results

• In 60% or more of posts: must-haves. They go in your headline, About section and top skills.
• In 25–60%: strong signals. Put them in your experience bullets.
• Missing must-haves: your learning list. Build a small project that uses each one so you can claim it honestly.
• Use their exact title wording. If posts say "AI Engineer" and you call yourself "GenAI Developer", you won't appear in their search.

Repeat the whole step for your backup role, so the profile shows up in both searches.

What good looks like

AI Engineer | LLM apps, RAG, Python | Built a support bot that answers 2,000 tickets/month

Title first (that's what search matches), then keywords, then one number that proves it. The numbers must be yours and true.

Put it on LinkedIn

1. Read profile.md line by line. Delete or fix every [CHECK] and anything not 100% true.
2. Let Claude do the typing:

   Open my LinkedIn profile edit pages and paste in profile.md one section at a time. Stop before each Save so I can review it.

3. Turn on Open to Work → Recruiters only. Enter your exact target titles and locations. Only recruiters see it. LinkedIn tries to hide it from recruiters at your current company, but can't guarantee that.
4. Fill Featured with 3 items: a short demo video, a GitHub repo or live project, and a one-page case study.
5. Set your location to the city you want to work in, and add certifications and courses from step 03.

1. Take 10 phone photos: facing a window for light, camera at eye level, plain wall behind you, neck and shoulders up. Put them in a folder called photos.
2. Let Claude pick and crop:

   Look at the photos in /photos. Rank them for a LinkedIn headshot on lighting, sharpness, eye contact and framing. Crop the best one to a 400×400 square with my face filling about 60% and save it as headshot.jpg.

3. Optional: run your best shot through an AI headshot tool for a cleaner background. It must still look exactly like you on a video call.
4. Banner:

   Design a 1584×396 LinkedIn banner image: dark background, my target title, and one proof point from profile.md. Save as banner.png.

After 500, LinkedIn just shows "500+", and every new connection extends your network reach, which affects who finds you in search. Aim for the right 500: people in AI hiring.

1. Import your contacts (phone and email) first. That's the easiest volume.
2. Have Claude build a target list:

   Search LinkedIn for technical recruiters and AI hiring managers at the companies in targets.txt. Save name, title, company and profile URL to leads.csv. Do not send any invitations.

3. You send 15–25 invitations a day, each with a short personal note. Claude can draft the notes.
4. Accept every relevant request. Leave thoughtful comments on recruiters' and AI leaders' posts daily; people connect with whoever keeps showing up.

• Ship one public project in your target role: a RAG chatbot on real documents, an agent that automates a real task, a model evaluation write-up. Link it everywhere. Recruiters forward proof to hiring managers; they don't forward claims.
• Post 3 times a week. What you built, a short screen recording, what broke and how you fixed it.
• Keep keywords in your headline. It shows next to your name in search, comments and invitations.
• Claim a custom URL like linkedin.com/in/yourname-ai and put it on your résumé.

• Numbers everywhere. "Cut support tickets 40%" beats "worked on AI".
• A 60-second demo video pinned in Featured. It's the easiest thing for a recruiter to forward.
• Reply within hours, with your availability and a link to the demo.

Tailor every application

Practice before the call

About an hour a day. Claude does the heavy lifting; you make the decisions.

1. week 1 · set up & research

   • Install Claude Code and the Chrome extension
   • Run the role-fit prompt, pick 2 target roles
   • Collect 25 job posts per role
   • Build keywords.md and your gap list

2. week 2 · rebuild the profile

   • Generate profile.md and fact-check it
   • Publish headline, About, roles, skills
   • New headshot and banner
   • Turn on Open to Work (Recruiters only)

3. week 3 · build proof

   • Ship 1 public project in your target role
   • Record a 60-second demo
   • Pin demo, repo, case study to Featured
   • Publish your first 3 posts

4. week 4 · reach out

   • 15–25 targeted invitations a day toward 500+
   • Comment daily on recruiter and AI posts
   • Tailor 5 applications a week
   • Mock interviews for every call you book

Your folder after 30 days

~/ai-job-hunt/
  resume.pdf      role-fit.md     postings.csv    keywords.md
  profile.md      headshot.jpg    banner.png      targets.txt
  leads.csv       photos/         project/

Re-run steps 04 and 05 every month. Job posts change, and your profile should change with them.