Overview
What this challenge is about.
You receive 40,000 anonymized de-identified intake-form records with two labels: urgency tier (4 classes) and routed sub-specialty (12 classes). Train (1) two independent classifiers (separate gradient-boosting and separate small Transformer), (2) a multi-task small Transformer with shared encoder + two task heads. Compare on macro-F1 per task, calibration, and total training time. Discuss positive/negative transfer between tasks and recommend a setup for the next clinical-pilot release.
The Brief
What you'll do, and what you'll demonstrate.
Decide whether a multi-task Transformer beats two single-task models on a healthcare triage benchmark, and characterize positive vs. negative transfer.
Earning criteria — what you'll demonstrate
- Apply multi-task learning with shared-encoder + task-head architecture
- Quantify positive and negative transfer between related tasks
- Evaluate clinical-grade classification with calibration in mind
- Recommend a deployment setup with clinical-leadership-readable framing
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
ML Researcher
Designing and characterizing multi-task transfer studies is the ML-researcher's signature deliverable on any clinical or scientific ML team.
This challenge sharpens
- multi-task-learning
- transfer-learning
- transformer
Applied AI Scientist
Pairing transfer analysis with clinical-leadership-readable framing is the applied-AI-scientist's daily craft at any healthtech AI company.
This challenge sharpens
- multi-task-learning
- model-calibration
- model-evaluation
Machine Learning Engineer
Shipping a shared-encoder multi-task setup with proper loss balancing is core MLE territory for any production triage system.
This challenge sharpens
- pytorch
- transformer
- multi-task-learning