About the Lab & Mission We are a stealth-stage applied AI research lab located in New York City, backed by top-tier investors and AI leaders. We are tackling one of the final frontiers in artificial intelligence: computational creativity. We are building a foundational intelligence layer engineered for superhuman creative reasoning, artistic decision-making, and divergent thinking. Our goal is to give anyone the ability to feel like a genius creator by simply chatting with AI. We operate as a tight-knit, high-velocity team of researchers, systems engineers, and world-class artists.
The Role:
We are seeking an ML Research Scientist to invent new methods beyond standard pattern matching and teach LLMs genuine creative reasoning, taste, and judgment.
What You’ll Do:
Invent Next-Gen Training Recipes: Develop novel post-training algorithms, RL recipes (GRPO, RLVR, DPO), and process-level reward verifiers to teach LLMs creative reasoning and long-horizon planning.
Master LLM Alignment & Reasoning: Advance Chain-of-Thought (CoT), System 2 reasoning, synthetic preference data generation, and reward modeling to guide model behavior in subjective, high-taste creative domains.
Unbounded Compute & Scale: Access massive multi-GPU clusters (H100/H200) to scale research ideas from zero-to-one concepts directly into production models.
Encode Creative Taste: Work side-by-side with world-class directors, designers, and writers to decode human creative judgment and embed it into model loss functions.
Ship Frontier Research: Move at early-stage startup speed with zero paper queues, middle management, or "cool demo" limbo—your research ships directly into systems used by millions.
What We’re Looking For:- Academic & Research Pedigree: PhD in CS/ML from a top institution (or an exceptional M.Sc. with a proven track record of breakthrough research).
- Publication Track Record: Multiple first-author papers at top-tier ML venues ( NeurIPS, ICML, ICLR
- ). Oral or Spotlight presentations are a strong positive signal.
- Deep LLM & Post-Training Expertise: Proven experience in LLM mid-training, continual pre-training, SFT, RLHF, GRPO, RLVR, or reasoning distillation.
- First-Principles Mathematical Depth: Mastery of probability, optimization, and statistical learning theory needed to invent new algorithms from scratch rather than wrapping existing frameworks.
- Founding DNA: High ambition, comfort with startup ambiguity, and a strong drive to own a significant piece of a technology revolution.
Why Join Us?
True Founding Ownership: $250k–$600k cash + 1.5%–2.5% founding equity ("Better to own 1% of a revolution than 0.001% of an evolution" ).
No Research Shelf-Life: Direct pipeline from whiteboard to production.
Frictionless Visa Support: Dedicated legal team providing fast-tracked O-1 / EB-1 petitioning and H-1B support.
Remote: across the US or in-person in NYC.