Overview
What this challenge is about.
You receive a labeled subset of an arrhythmia ECG dataset (about 80,000 10-second windows, 4 classes), a microcontroller latency lookup table (op-level milliseconds) for a Cortex-M7 target, and a starter Once-for-All-style supernet. Run a small evolutionary or random search over the supernet conditioned on a latency budget of 50 ms per inference. Evaluate the top 5 candidates on F1 per class, latency on the lookup table, and parameter count. Recommend one model and write a 2-page memo on whether to invest in a larger NAS pipeline.
The Brief
What you'll do, and what you'll demonstrate.
Run a small hardware-aware NAS for a microcontroller ECG classifier and decide whether a full NAS investment is justified for the next product cycle.
Earning criteria — what you'll demonstrate
- Define a hardware-aware NAS search space conditioned on a latency budget
- Run and evaluate a small evolutionary/random NAS
- Use latency lookup tables to avoid full-device measurement during search
- Reason about the ROI of NAS at a startup's compute scale
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.
Applied AI Scientist
Running a small NAS pipeline and translating the results into a scale-up recommendation is exactly the day-one work of an applied AI scientist at a healthtech or edge-ML startup.
This challenge sharpens
- neural-architecture-search
- hardware-aware-design
- model-optimization
Machine Learning Engineer
Hardware-aware model design with latency budgets is the MLE craft of shipping ML where it actually has to run.
This challenge sharpens
- hardware-aware-design
- edge-inference
- model-optimization
ML Researcher
Designing the search space and the proxy-vs-true latency validation is the kind of methodology question ML researchers tackle in industry research teams.
This challenge sharpens
- neural-architecture-search
- evaluation
- pytorch