Overview
What this challenge is about.
Use the publicly-available PCB defect dataset (e.g., DeepPCB or HRIPCB). Fine-tune a small object detector (YOLOv8n or RT-DETR-small) on the 6 defect classes. Evaluate mean Average Precision (mAP) overall + per-class, plus a hard false-positive budget (fewer than 1 FP per 100 boards) at recall above 90 percent. Deliver the trained model, evaluation report, model card, and a 3-page line-acceptance memo for the QA team.
The Brief
What you'll do, and what you'll demonstrate.
Train a PCB defect detector that achieves recall above 90 percent at the FP budget across the 6 defect classes.
Earning criteria — what you'll demonstrate
- Fine-tune a modern object detector on a defect-detection task
- Operate under a hard false-positive budget
- Write a model card that supports a real production decision
- Communicate model boundaries to a non-ML QA team
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.
Machine Learning Engineer
Defect-detection models with a hard FP budget are the day-one MLE work at any manufacturing-AI or industrial-vision company.
This challenge sharpens
- object-detection
- transfer-learning
- ml-pipelines
Computer Vision Engineer
Defect-detection on real PCBs builds the CV-engineer judgment around class imbalance, augmentation, and per-class evaluation.
This challenge sharpens
- object-detection
- data-augmentation
- model-evaluation
AI Safety Researcher
Operating under explicit FP budgets and writing honest model cards is the safety-aware engineering pattern AI safety researchers practice.
This challenge sharpens
- model-evaluation
- object-detection
- transfer-learning