Ship a Lightweight ML Microservice for an EdTech Reading App
Overview
What this challenge is about.
You receive 3 months of session telemetry (around 50M reading events, child-anonymized). Engineer features per session window, train a small classifier (logistic regression baseline + LightGBM challenger), and package the chosen model behind a FastAPI service that responds within a 200 ms p95 latency budget on a CPU-only container. Provide load-test results, a model card, and a Dockerfile the mobile team can hand to platform.
The Brief
What you'll do, and what you'll demonstrate.
Train and deploy a sub-200ms reading-pace anomaly microservice that the mobile team can integrate this sprint.
Earning criteria — what you'll demonstrate
- Engineer per-session features under tight latency constraints
- Compare a simple baseline against a stronger model honestly
- Package and load-test a model as a real microservice
- Write integration docs a downstream team can actually use
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.
MLOps Engineer
Shipping a containerized, load-tested inference service with integration docs is the day-one MLOps engineer work at any product-AI company.
This challenge sharpens
- model-serving
- containerization
- ml-pipelines
Machine Learning Engineer
Closing the loop from model training to serving with a latency budget is core MLE work, especially at mobile-first edtech companies.
This challenge sharpens
- feature-engineering
- model-serving
- ml-pipelines
AI Engineer
AI engineers regularly own the boundary between an ML model and a product team — this challenge practices that boundary end-to-end.
This challenge sharpens
- model-serving
- containerization
- model-evaluation