Senior/Staff Applied Scientist / Research Engineer

About The Job
The client is hiring for this role. Harper is an AI career agent that works for you — the candidate. Skip the application form: just talk to Harper, and it puts strong candidates straight in front of the client — plus other frontier roles worth your time.

Job Title
Senior/Staff Applied Scientist / Research Engineer

Salary
Competitive base + Equity — top-tier comp at a world-leading AI lab

Location
Seoul (Hybrid)

Company Description
An Nvidia-backed European AI lab, valued at ~$14B — one of the world's most important, building open-weight, open-source frontier models that go head-to-head with the best closed labs. They train the models, not wrap them. Teams across France, the US, UK, Germany, and Singapore — now building in Seoul. Low-ego, ship-fast, research-meets-production.

What You'll Do

• Run pre-training, post-training, and deployment of SOTA models on clusters with thousands of GPUs — you don't panic at OOM errors or when NCCL refuses to talk.
• Build SOTA models across modalities — text, image, speech — and invent the methods to push them further.
• Generate and curate data for pre/post-training, own the evals, and make the model beat expectations.
• Build the tooling and frameworks for data generation, training, evaluation, and deployment.
• Tackle complex use cases with agents and RAG pipelines, and run research projects directly with client research teams.

Who You Are

• Expert with PyTorch or JAX; you write clean, high-performance, fault-tolerant Python.
• You dive into a huge codebase and find your way with little guidance.
• You don't need a roadmap — you just do. You don't need a manager — you just ship.
• Low-ego, collaborative, and always learning.
• A track record of success — personal projects, industry, or academia.
• Fluent in English and Korean.
• Great if you have: a PhD/Master's in a relevant field (Math, Physics, ML) — or you're exceptional from another background; research range across agents, multimodality, robotics, diffusion, or time-series; contributions to a large, widely-used codebase; top-venue publications.

Why Join

• Train frontier, open-weight models on serious compute — thousands of GPUs, real research freedom.
• Publish, open-source, and ship — research that actually reaches production.
• Work with a world-class global research org, from Seoul.
• Low-ego, ship-fast culture. Competitive comp + equity.