Lilly Hackathon 2026

Challenge 1 · ForecastIQ – Predict smarter [Finance & Commercial Analytics]

A medicines company has to guess how much product its distributors will order next month. Guess too low and patients miss out; guess too high and money sits in a warehouse. Today that guess relies on last year’s numbers and an experienced person’s memory that “June is always strange” or “that distributor orders every second Tuesday”. You’ll get three years of weekly order data for a made-up product, plus a calendar of the events that throw orders off: public holidays, end of financial year, big contract renewals, promotions. Build a tool or model that uses those events to produce a better forecast, and explains in plain English what is driving it. Great for teams who like data, numbers and clear visuals. 

Challenge 2 · PulseScore – Cut through the noise [Commercial Operations & Analytics]

Jordan looks after 120 doctors and clinics across Western Sydney. Every Monday, Jordan spends 45 minutes checking four different systems (visits, emails opened, webinars attended, articles downloaded) and still ends up planning the week on gut feel. One doctor has been reading everything the company sends and hasn’t had a visit in six weeks, because no single screen shows the whole picture. You’ll get 90 days of activity data for 80 made-up doctors in four linked files. Build PulseScore: one simple score per doctor that shows who is most engaged, which “hidden gems” are being missed, and what is behind each score. You’ll need to explain why you weighted things the way you did. Great for teams who mix analytics with business sense and a good eye for dashboards. 

Challenge 3 · PrivacyAssurance – Protect what matters. [Privacy, Compliance & Risk]

When someone calls a medicines company with a question, the call is logged in a record. Part of the record allows limited patient details; the free-text notes do not, because those notes are shared with other teams. Writing “patient MH, female, born 03/1962” in the right box is fine. Writing “Margaret Hollingsworth called back” in the notes is a privacy breach under Australian and New Zealand law, and it happens by accident when someone types quickly. With thousands of records a month, the reviewer can only check a handful. You’ll get 25 sample records containing real breaches, near-misses (is that a pharmacy name or a person’s name?) and some genuinely tricky cases. Build a checker the reviewer can trust, then decide what should happen when it finds something: flag it, remove it, notify someone, or stop the record being closed. Great for teams who enjoy clear thinking about rules, judgement calls and how people and tools work together.