You are viewing a preview of this job. Log in or register to view more details about this job.

AI Systems Intern

The Autoimmune Fund
Email

About The Autoimmune Fund

Autoimmune and immune-mediated diseases affect millions of people, most of them women. Yet the money has gone almost entirely one way. Of the $40.0 billion in venture funding that went to the field from 2014 to 2025, 94.4% went to drug therapeutics. Prediction, diagnostics, devices, care access, monitoring and remission maintenance shared the remaining 5.6%. The Autoimmune Fund, a seed and Series A venture fund dedicated exclusively to autoimmune and immune-mediated disease, exists to back the rest.

We invest across the whole spectrum of care. We lead investments everywhere except drug therapeutics, where we co-invest. Our thesis follows three scientific revolutions reshaping the field: Immune Precision, Accelerated Aging, and Female Biology.

We are a small team that runs on AI. We use it to source companies, analyze markets, and produce our research, including The Autoimmune Decade, our annual report on the field, launching in January 2027. That is why this role matters: the systems you build will shape how the whole fund works.

 

What You'll Do

You'll start with two projects that set the standard for everything we publish, then take on AI work across the whole fund.

First project: align our visual communication system. Audit our palette, typography, icons, chart styles and layout rules across documents and slides. Find where they conflict and resolve them. Turn the result into one clear, written source of truth.

Second project: train Claude on our design system. Set up our brand inside Claude's design tools (Claude Design) and write the instructions and reusable skills that teach it our rules. Generate slides, one-pagers and infographics, compare them against the standard, find where the AI drifts, and fix the instructions until it stops.

Then: build AI workflows for a small, fast team. Turn the fund's recurring work into reliable AI tools, skills and guides, so everyone can produce high-quality work without starting from scratch.

 

You Have

  • Current enrollment in an undergraduate program (design, cognitive science, computer science, communications, or the life sciences)
  • Real, hands-on AI use. You've built prompts, custom instructions, projects or workflows in tools like Claude or ChatGPT, and you can show us one
  • A strong eye for design systems: type, color, grids, hierarchy, consistency
  • Clear writing. Training an AI is mostly writing precise, testable instructions
  • Comfort in Figma or a similar design tool
  • Obsessive attention to detail. You notice when a teal is one shade off
  • Availability to work in person in Los Angeles

 

Nice to Have

  • Coursework or lab experience in immunology, biology or another life science
  • Basic HTML/CSS or Markdown
  • Experience building or maintaining a brand guide or style guide
  • Interest in healthcare, biotech or venture capital

 

What You'll Get

  • Direct mentorship. You work side by side with the fund's Managing Partner, a clinician and former Wall Street healthcare equity analyst who covered 500+ companies over 14 years.
  • Work that ships. What you build goes into real investor materials and our annual report, not a drawer.
  • Builder-level AI skills. You will design, test and fix AI systems, not just prompt them.
  • A front-row seat to venture. See how a specialist fund sources companies, weighs science and makes decisions.
  • A portfolio piece few undergrads have. Working AI systems you built and trained.

 

To Apply

Send your resume, a portfolio or work samples, and a short note (200 words max) describing one thing you've built or automated with AI and what you learned when it went wrong to bonnie@autoimmune.fund. Then we will send you the required information for the work test below.

 

Work Test (3 to 4 hours, unpaid)

We'll send you a packet: an excerpt of our brand rules, two pages of raw content (Content A and Content B), and one sample AI output. Send everything back as one PDF plus your instruction file. Unfinished parts are fine if you tell us what you would do next.

1. Build it. Write one set of instructions (a skill, custom instructions or a system prompt) that turns raw content into a one-page summary following our rules. The same instructions must work on both Content A and Content B. No hand edits to the AI output.

2. Break it. Write a third input of your own that makes your instructions fail. Show the failure, then show the fix.

3. Grade our sample. The sample AI output in the packet has problems. List every one you find and the rule each breaks. Flag anything else in the packet you think is wrong.

4. Score yourself. Write at least ten pass/fail checks for this task. Score your own outputs against them. An honest 7 out of 10 beats an inflated 10 out of 10.

5. Show the work. Share the full chat logs. In 150 words or fewer, tell us one thing in this task you would never let the AI do unsupervised, and why.