Open role

Research Scientist InternApplied ML for Creative Tooling

Playful Design is the designer in your pocket. AI-native creative tooling used by serious social creators. We ship fast, care about taste, and want a research scientist intern who turns computer vision and generative model research into features people actually use. You'll be training, integrating, and evaluating models that ship straight into the product.

Work authorization

This is an internship, so we don't sponsor work visas for it. You'll either need existing U.S. work authorization — F-1 students on CPT/OPT are welcome — or you can work with us fully remotely from somewhere you're already authorized to work. If you have a special situation that lets you work with us another way, mention it and apply anyway.

What you'll do

Turn computer vision and generative model research into features people use. Train models, build ensembles, and ship the results.
Build image understanding that gives our creative tools a designer's eye, and mine signal from real creator usage to make generation and suggestions smarter.
Design evaluations that tell us which models and approaches actually work, and hunt down the strange, useful models in the image and design literature worth turning into product.

What we're looking for

Deep in computer vision, multimodal models, and segmentation, plus the image understanding literature.
At least one CVPR-tier paper, or time as a research engineer or an engineer embedded on a research team.
You've built real generative or creative ML products (think Adobe Firefly), inventing patentable tech.

Apply

Full name *

Email address *

City, State, Country *

One line is fine (e.g. San Francisco, CA, United States).

Remote, hybrid, or in-person preference *

Résumé or portfolio link *

One link is enough (Google Drive, Dropbox, personal site, etc.). Use whichever best represents you.

Work authorization / visa status *

Will you need immigration or visa support for this role? *

Why this role? *

For each area below, pick 1 to 5 using the descriptions under each question.

Computer vision & multimodal: image understanding, segmentation, VLMs *

1: No experience
2: Coursework / tutorials; used pretrained CV/VLM models
3: Ships CV/multimodal features; fine-tuning, segmentation, detection
4: Advanced; custom architectures, multimodal pipelines, strong results
5: Expert; novel CV/multimodal research, deep in the literature

Generative models: training, fine-tuning, building (not just calling APIs) *

1: Used generation tools / APIs only
2: Basic fine-tuning / LoRA; toy generative projects
3: Ships generative features; fine-tunes, builds ensembles, controls quality
4: Trains/customizes models; diffusion/transformer internals, pipelines, evals
5: Architect; novel generative methods, training at scale, research-grade

Turning research into shipped product: evaluation, model routing, preference / recsys, data mining *

1: No experience
2: Ran experiments / basic metrics
3: Builds evals; ships research-backed features to users
4: Designs eval systems; routing, preference models, data mining at scale
5: Expert; owns applied-research strategy end-to-end (research to prod)

When do you feel most playful in your work? *

When was the last time you took a risk that paid off? *

What are you open to? (select all that apply) *

Part-timeFull-time

Date range you're available *

From

Through

Anything else? (optional)