Rooms, Reimagined

Summery

Innovate

In 2024, I joined the Target Creative team as an AI Fellow. My job was to research and prototype new ideas with Generative AI, fast. The capstone project? Build a system that lets Target customers see how Target’s home goods could could change their spaces. I worked with a colleague to develop this project. We delivered a working prototype pitch to the VP of marketing.

⤢Enlarge

⤢Enlarge

⤢Enlarge

⤢Enlarge

⤢Enlarge

Some images of prototypes & presentations I delivered.

Close ×

Close ×

Close ×

Close ×

Close ×

The Challenge

Operate like a startup inside a Fortune 500.

In 2024, a colleague and I joined Target’s creative division as its first-ever AI fellows. The mandate: research models deeply, test what they could really do, and find business applications worth building. Target already had a 3D experience where you place furniture models into a virtual room. The capstone brief asked for something radically beyond it.

The Brief

Take a photo of your room. See it restyled with products you can actually buy.

Could a system read a customer’s own space, pull matching products from Target’s catalog, and generate an image of that room furnished with them, with a path to purchase?

The Approach

In 2024, models couldn’t just do this. We had to make them.

• Six weeks testing multiple models against capability, speed, and cost
• Built a pipeline that decomposed the user’s photo into versions the model could understand, so it restyled their room instead of hallucinating a new one
• My lane was the tech: the Python image-processing scripts and the wiring of the entire workflow, end to end

The Outcome

A working prototype, presented to the VP.

Photograph a room, pick a style from Target’s catalog, get it back reimagined with purchasable products. It was a proof of concept, not a shippable product, and we said so. What we proved was the workflow, a year before this became a few-prompts problem.