Overview
What this challenge is about.
Build a visual search prototype for sustainable fashion using CNNs and FAISS, evaluate retrieval quality, and earn a verifiable certificate.
The scenario
The client is a mid-sized direct-to-consumer online retailer specializing in ethically sourced and recycled-material clothing, operating across Western Europe with a catalog of about 500,000 products and a small in-house data team. Search-driven sessions convert poorly because shoppers often cannot name a style, fabric, or cut they have seen elsewhere, leaving discovery as the company's biggest growth bottleneck.
The Brief
What you'll do, and what you'll demonstrate.
Prototype and justify a visual search system that lets shoppers find catalog items by uploading a photo rather than typing keywords.
Earning criteria — what you'll demonstrate
- Extract image embeddings from a pre-trained convolutional neural network without fine-tuning
- Build and query an approximate nearest-neighbor index with FAISS for fast similarity search
- Evaluate retrieval quality rigorously using precision at k and recall at k
- Translate model performance metrics into a defensible estimate of business impact
- Package a machine learning prototype so another engineer can reproduce it on a laptop
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Aligned coursework coming soon.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
- Deep Learning
Design and train neural networks for complex pattern recognition tasks.
- Computer Vision
Build systems that interpret and analyze visual information from images and video.
- Cnn
Apply cnn to solve real industry problems and demonstrate production-level capability.
- Faiss
Apply faiss to solve real industry problems and demonstrate production-level capability.
- Embedding
Apply embedding to solve real industry problems and demonstrate production-level capability.
- Prototyping
Apply prototyping to solve real industry problems and demonstrate production-level capability.
Careers
Career paths this challenge builds toward
Completing this challenge demonstrates skills that transfer directly to these roles:
Computer Vision Engineer
This challenge mirrors the core work of a computer vision engineer: turning raw images into useful features and building retrieval systems that power real product discovery. You practice the exact pipeline of embedding extraction and similarity search used in industry visual search.
This challenge sharpens
- computer-vision
- cnn
- embedding
Machine Learning Engineer
Prototyping a deep learning system end-to-end, evaluating it honestly, and justifying it to the business is the daily reality of an ML engineer. You leave with a portfolio piece that shows you can ship a working model and defend its value.
This challenge sharpens
- deep-learning
- faiss
- prototyping
Applied Research Engineer
Applied research roles reward engineers who can adapt pre-trained models to new domains and rigorously measure retrieval quality. This challenge builds that muscle by demanding clear metrics and reproducible experiments on a real fashion dataset.
This challenge sharpens
- deep-learning
- computer-vision
- embedding