Overview
What this challenge is about.
You receive synthetic wearable telemetry (heart rate, accelerometer, sleep stages) for around 5,000 patients across 90 days, plus the existing scattered feature scripts from the cardiology team. Define ten reusable features (e.g., resting heart rate seven-day rolling average) with explicit time-aware semantics. Implement offline materialization to a warehouse table and online serving from a low-latency store (Redis or DynamoDB). Provide a Python SDK with two calls (get_features_for_training, get_features_for_inference) and prove offline-online parity (less than 0.1% value drift) on a 1,000-patient sample.
The Brief
What you'll do, and what you'll demonstrate.
Build a feature-store slice that guarantees consistent feature values between offline training and online inference for at least ten production features.
Earning criteria — what you'll demonstrate
- Model time-aware features with explicit event-time semantics
- Implement an offline-online parity guarantee for ML features
- Design a small but real SDK ML teams actually want to adopt
- Reason about feature lineage and reproducibility
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.
Data Engineer
Building the data-platform layer that ML teams depend on is core data-engineer work at any ML-driven company; feature stores are a hot specialization.
This challenge sharpens
- data-modeling
- etl
- online-offline-parity
MLOps Engineer
Feature stores sit at the heart of MLOps; owning the parity guarantee is a top-three responsibility in most MLOps role descriptions.
This challenge sharpens
- feast
- online-offline-parity
- python
Machine Learning Engineer
Knowing the feature-store layer from the inside makes a junior MLE dramatically more effective at debugging production model regressions.
This challenge sharpens
- feature-engineering
- python
- feast