Research Scientist, Agents

Storm3; New York (NY); In loco; USD 250.000 - 350.000; Tempo pieno

Vantaggi offerti da questo lavoro

Founding-level equity

Descrizione del lavoro

Storm3 in New York invites a Founding Research Scientist to join a Scientific Superintelligence Lab (Stealth). You will lead research on agents that learn continually and publish at top venues. Equity accompanies a competitive salary in the $250k–$350k range, reflecting founding-stage ownership.

Full-time, in-person in New York. The role emphasizes independence to define a research direction, bridging hypothesis, experimentation and results in a frontier field of AI for science.

Competenze

• Published work at NeurIPS, ICML, ICLR, CoRL, CVPR or similar.
• Independence to own a research direction.

Mansioni

• Lead research on agents that learn continually from experience, without retraining and without forgetting.
• Develop learning methods (continual learning, memory, post-training, RL, reward and preference learning).
• Build computer-use and embodied agents that act in real time and learn from a single demonstration.
• Design evaluations that measure generalization, adaptation and scientific competence.
• Publish at top venues and help define the field.

Conoscenze

Continual learning research

Multimodal AI

Reinforcement learning

Formazione

Final-year PhD / Postdoc

Descrizione del lavoro

Founding Research Scientist
Scientific Superintelligence Lab (Stealth)
New York, NY
$250,000 – $350,000 Salary + Equity

Come join a research-led company in New York working toward scientific
superintelligence: AI that can drive science end to end, from hypothesis to experiment to result.

Responsibilities:

• Lead research on agents that learn continually from experience, without retraining and without forgetting.
• Develop learning methods (continual learning, memory, post-training, RL, reward and preference learning) for long, multimodal, partially observed experience with sparse and delayed outcomes.
• Build computer-use and embodied agents that think only when needed, act in real time and learn from a single demonstration.
• Design evaluations that measure generalization, adaptation and scientific competence rather than benchmark scores.
• Publish at top venues and help define the field.

Requirements:

• Published work at NeurIPS, ICML, ICLR, CoRL, CVPR or similar.
• Breadth across at least two of vision, language, RL, generative models and robotics;
• comfortable moving between them as the problem demands.
• Unafraid to try unconventional approaches to push beyond the frontier.
• Genuine curiosity about how experimental science works.
• Seniority: final-year PhD, postdoc, research scientist or senior research scientist, with
• the independence to own a research direction.

Nice to have

• Research on continual learning, long-range memory, RL or post-training for LLMs, VLMs or agents.
• Experience with computer-use, GUI, web or embodied agents.
• Prior work in AI for science.

Full-time, in person in New York; Boston-based candidates considered.
Founding-level equity and a competitive salary. Reports to the CSO.