Business Process & AI Operations Intern

Why this role exists

CarePool moves people to medical appointments — dialysis, chemotherapy, physical therapy. The appointments you cannot skip. What separates us from a rideshare app is the stakes: if an Uber is late you are annoyed, and if we are late somebody misses treatment.

That makes us an operations business with a real product, real constraints, and considerably more opportunity than we have people to chase. This role is how we chase some of it.

We have deliberately dropped the technical-background requirement. AI has collapsed the distance between understanding a problem and being able to build something about it. What has not collapsed — what we think has actually become more valuable — is judgment: choosing the problem, directing the tools, telling when a confident answer is wrong, and getting something across the finish line.

The two halves of the job

We want to be straight about both of them, because only one is interesting.

The second half is an apprenticeship. You work directly with our CTO and leadership team, learning to do what they do: point AI at a messy business problem, and come back with something useful. Over time that means auditing how a department actually works, finding where automation earns its keep, designing the workflow, and seeing it through to people using it.

What you will actually do

• Get drivers onboarded into our automated assignment program — the near-term job, start to finish.
• Sit with people in operations, dispatch, support and finance and work out where the manual effort really goes.
• Interview riders, drivers and partners about what is not working, and turn what you hear into something actionable.
• Build and test AI-assisted workflows for the work you find, then judge honestly whether they were worth it.
• Put dates, costs and owners on your proposals, so they can be decided on rather than admired.
• Track initiatives from idea to live, including the unglamorous middle where most of them die.
• Document what you build well enough that the next person can run it without you.
• Train internal teams as new capability lands — adoption is part of the work, not a follow-up to it.

Who we are looking for

We care much less about your major than about how you think. Entrepreneurship, business, management, communications, information systems, operations, economics, marketing and industrial engineering are all common backgrounds here — and none of them is a requirement.

You will do well here if you are the sort of person who asks why we are doing it this way, whether there is a smarter workflow, and what you would change if this were your company. You look for the opportunity rather than waiting to be handed one.

• You are a current undergraduate or graduate student.
• You can talk to technical and non-technical people without changing what you actually mean.
• You are comfortable working independently and telling us early when you are stuck.
• You use AI tools seriously — ChatGPT, Claude, Gemini, Copilot — and you check what they tell you.
• You can say "I do not know" cleanly, which is rarer and more useful than it sounds.

How we hire for this role

We do not interview for this one. An interview cannot detect whether you catch an AI being wrong — but a real project with a live Q&A can, because the failure mode is obvious within about four minutes. So we run The CarePool Challenge, and we pay you to take part.

1. A fifteen-minute phone screen. Three questions, the same three for everyone. Mostly: tell us about something you worked out using AI recently, and when it told you something wrong.
2. Six finalists are invited. You get a materials pack — our product documentation, a business primer, our operating numbers — and a login to an interactive demo of the whole product: admin, org admin, driver and passenger. All of it is cleared for you to use with any AI tool you like.
3. About ten days, 15–20 hours. Use AI to audit our business and bring us one recommendation: a new product, a new service, or an improvement to something we already do — with an execution plan real enough that we could act on it.
4. An hour-long presentation, in person. Twenty minutes on your audit and recommendation, fifteen on our questions, fifteen on driver onboarding, and time at each end. We will interrupt, and we will push on your numbers.

What the Challenge asks for

Your idea does not have to be one we would build. It can be wildly off-base. We are not buying your idea — we are watching how you got to it. A brilliantly-reasoned recommendation we would never pursue beats a safe, obvious one you cannot defend.

Deliver it in whatever form you want — slides, a video, a PDF, a demo site, a working prototype. Format is not scored. It has to cover five things:

• Your methodology. Which tools, which prompts, how many iterations, what failed, where you went outside the AI. Inside this, one required section: at least three specific instances where an AI told you something wrong and how you caught it. Quote the wrong output. This is the single most important part of your submission.
• What you learned. Your read on the business — what surprised you, what did not add up, and what you wish we had given you.
• Your recommendation. One primary idea, developed properly, plus up to two you considered and set aside, and why.
• An execution plan. Dates, costs, resources and lift. Name your assumptions — a plan that says "this only works if trip volume exceeds X" is stronger than one that quietly assumes it. If anyone has to adopt your idea, add a page on how you would get them to.
• A driver-onboarding plan. One page on the real near-term job: how would you get forty drivers to enter their weekly availability within two weeks?

How we score it

What loses points: an AI-generated deck you cannot explain, confident numbers with no source, a recommendation that ignores what the business is, and a plan with no dates or costs. What wins: evidence you checked things, noticing something we did not tell you, and saying "I do not know" cleanly when you do not.

Use AI as much as you want — that is the point, and there is no penalty for AI-generated work. Do the work yourself, though: using AI is expected, having another person do it for you is not, and the Q&A is very good at telling the difference. Your ideas remain yours; taking part does not transfer ownership of your concept to CarePool.

What you get out of it

• Paid, flexible around your class schedule, with the possibility of long-term employment.
• Direct mentorship from our founders and CTO — not a project handed down through three layers.
• Real experience at the intersection of AI, operations and business strategy, in a company small enough that you can see the whole thing.
• A body of work you can actually show people: an audit, a recommendation, an execution plan, and something you shipped.

Our goal is not to produce interns. Within a few years of graduating we would like you to have the skills and the confidence to build your own company, launch your own AI-powered venture, or lead innovation wherever you end up.

Eligibility

• This role requires US work authorization. It is open to candidates on CPT and OPT.
• Junior, senior or graduate standing, working toward a bachelor's degree or beyond.
• Do not contact our customers, drivers, riders or staff for research during the Challenge — ask us instead.