Applied AI for Healthcare and Pharma

Rohit Kelapure

Co-founder & FDE @ 8090 · Healthcare AI

Your healthcare AI demo works. Now lead it through the review.

Building the demo is the easy part. Then the review starts.

What happens when it's wrong?

Who catches it?

How do you know it's good enough?

And connecting it to the systems you already run takes longer than building the AI.

The engineer who can answer those questions leads the next project.

Over four Saturdays you take one real use case in patient intake, claims, medical documents, or quality records to an approved pilot. Each week builds part of a review packet: risk screen, review-gate design, eval harness, legacy-rule tests, and decision memo. On the fifth Saturday you defend it to a mock review board. Bring your own case or use a public or synthetic one. No real patient data is used in class.

I teach what I do on-site at 8090, embedded with the people who do the work. One on-site discovery at a medical device manufacturer ran 31 days before anyone wrote a line of code. On a federal Medicare claims program, our pipeline extracted about 100K rules in about 40 days. Across six applied AI projects in 2025, one fax-intake system files 82% of incoming faxes automatically. Another delivered 33% quality against an 80% target and needed a fix-it sprint. You get both stories.

What you’ll learn

Take one healthcare or pharma AI use case from idea to an approved pilot, one Saturday at a time.

  • Score volume, staff-hours at stake, the cost of a wrong answer, integration path, and which rules apply on one risk screen.

  • Write the intended use and stop criteria, so a weak idea dies in week one instead of month six.

  • Shadow the people doing the work and log every exception path, with an owner for each one.

  • Record today's baseline: volume, error rate, and handling time, so the pilot has something to beat.

  • Route low-confidence work to trained reviewers, map every hop PHI touches, and keep an audit trail a reviewer can follow.

  • Seed known-wrong items into the queue and track catch and override rates, so review doesn't become a rubber stamp.

  • Build a golden set from your exception log, with messy records, red-team cases, and subgroup slices.

  • Score omissions as well as hallucinations, check the judge against human labels, and run it on a vendor tool too.

  • Use AI to extract business rules from legacy claims, EHR, and lab code, then verify each with the people who run it.

  • Turn verified rules into tests, so AI-assisted changes to the old system are checked against what it does today.

  • Write the decision memo: the business case in staff-hours and rework, measured against today's error rate.

  • Set tolerances, drift checks, a model-change rule, and a rollback owner, then defend it at a mock review board.

Learn directly from Rohit

Rohit Kelapure

Rohit Kelapure

Co-founder & FDE @ 8090

See all products from Rohit

Who this course is for

  • Engineers and Developers in healthcare, insurance, medtech, or pharma asked to ship AI where a wrong answer risks patients or compliance.

  • Forward-deployed and solutions engineers running AI projects for healthcare or pharma clients who want a repeatable review method.

  • Engineers moving into healthcare who want a portfolio piece. Product, quality, and validation leads are welcome too. Labs have templates.

What's included

Rohit Kelapure

Live sessions

Learn directly from Rohit Kelapure in a real-time, interactive format.

4 live Saturday sessions plus a demo day, all recorded

Four Saturday sessions (teaching + lab) plus demo day. Every session is recorded.

Office hours after every live session, about 2.5 hours in total

12:30–1:00 PM PT after every live session.

The Review Packet playbook: 10 templates, from risk screen to decision memo

Opportunity brief, exception log, review-gate design, golden-set and judge starter, validation evidence checklist, decision memo, pilot plan.

Eval harness starter repo with golden-set schema and fail-closed judge

Starter repo for your own golden set and judge, with the golden-set schema included.

Public and synthetic lab cases, no real patient data

No real patient data is used in class.

Mock review board with written feedback on your pitch

Written feedback on every pitch.

Free 90-minute 1:1 with Rohit, used by May 7, 2027

Bring one healthcare or pharma AI project and leave with a go or no-go read, the risks a reviewer will raise first, and your next step. One per student, used by May 7, 2027. Book your slot: https://calendar.app.google/ieRhVAeFXyG4StSDA

Free re-take of a future cohort

Join a future cohort at no cost.

Lifetime access to recordings and materials

Go back to every recording, template, and lab whenever you need to.

Maven Guarantee

Your purchase is backed by the Maven Guarantee.

Course syllabus

Week 1

Jan 2—Jan 3

    Setup week (async), Sat Jan 2 – Fri Jan 8, 2027

    1 item

Week 2

Jan 4—Jan 10

    Jan

    9

    Session 1 · Pick the problem worth solving

    Sat 1/91:30 PM—3:30 PM (EST)

Frequently asked questions

Maven for Teams

Reimbursement

Get your company to pay

Everything L&D needs: email template, receipts, and certificate of completion.

Get reimbursed

Team discount

Learn with your teammates

Save 20%+ when 2 or more teammates enroll in the same cohort.

Save 20%+ with a team

Private cohort

Run a cohort for your org

A dedicated cohort with a custom schedule and curriculum, tailored to your team.

Book a private cohort

$1,800

USD

Jan 2—Feb 6
Enroll