Overview
What this challenge is about.
You will design a feature-store layer covering 12 representative fraud features (account-level, merchant-level, transaction-level), with both batch (Spark) and online (low-latency) read paths. Prototype using Feast or a Postgres-based reference implementation. Demonstrate train-serving consistency by computing the same feature offline and online on a 1M-row sample, with consistency within 0.1 percent. Deliver the architecture doc, prototype, and a 2-page rollout plan covering the remaining 28 features.
The Brief
What you'll do, and what you'll demonstrate.
Design and prototype a feature store that guarantees train-serving consistency for fraud features at SME-payments scale.
Earning criteria — what you'll demonstrate
- Design a feature store covering batch + online read paths
- Demonstrate train-serving consistency as a measurable property
- Choose between off-the-shelf and bespoke for a real ML platform decision
- Write a rollout plan that engineering can execute
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.
Career paths this builds toward
Canonical rolesData Engineer
Designing and shipping a feature store with documented consistency is core senior-data-engineer work at any ML-heavy fintech.
This challenge sharpens
- feature-stores
- data-pipelines
- system-design
Machine Learning Engineer
Owning train-serving consistency is the bread and butter of ML engineering at companies running production models against money movement.
This challenge sharpens
- feature-stores
- train-serving-consistency
- python
MLOps Engineer
Feature-store platform work bridges directly into MLOps territory, especially around monitoring and contract enforcement.
This challenge sharpens
- feature-stores
- system-design
- data-pipelines