Intern

Egra•New York City, NY

13h•Onsite

About The Position

This posting is for a role with wide possibilities, including full-time researchers, full-time engineers, and interns. The expectation is that individuals will have complete ownership over their work from day one, without lengthy onboarding, waiting for permission, or navigating layers of approval. The work done in the first month will be integrated into the product. The company is training models on high-bandwidth data such as biosignals, eye-tracking, pupillometry, video of human faces, audio, and the content humans are reacting to, along with downstream behavior. The company is building toward a future where AI is optimized for its impact on the human user.

Requirements

Ability to take a vague research direction and ship something concrete within a week
Strong opinions about what makes representations actually generalize, informed by personal experiments
Comfort with heterogeneous, multimodal data and a toolkit for making it useful
Ability to ship fast with AI coding tools (Codex, Claude Code, agents)
Taste: ability to look at a benchmark and identify why it's leaky, or at a model architecture and identify what it can't possibly learn
General modeling ability on hard data

Nice To Haves

PhD, no PhD, dropped out, finishing undergrad are all fine.
Interns who out-shipped postdocs have been hired and will be again.

Responsibilities

Designing self-supervised pretraining objectives on multimodal physiological + content data
Stress-testing recent multimodal / signal foundation model papers to understand exactly where and why they break under distribution shift
Building evaluation protocols that distinguish real progress from leaky-benchmark noise
Shipping the internal research tooling (experiment tracking, dataset versioning, agentic eval pipelines) that lets research compound instead of repeat
Closing the loop between offline model results and the live product
Writing internal research memos that become the shared knowledge base