Odyssey is looking for a researcher to pioneer interactive video models capable of real-time generation, understanding, and interactivity, aiming to redefine generative media for various applications.
Requirements
- deep expertise in machine learning and a track record of original contributions to generative modeling, video synthesis, multimodal learning, or related areas.
- Experienced in conducting long-horizon, exploratory research that produces conceptual and algorithmic breakthroughs beyond scaling improvements alone.
- Comfortable working from first principles: defining new problems, designing experiments to test them, and building minimal systems to prove what’s possible.
- Motivated by understanding as much as by performance: you care about how and why models work, not only that they do.
Responsibilities
- Pursue original research into the foundations of generative interactive video : representation, temporal coherence, causality, control, and long-horizon imagination.
- Formulate and own an independent research agenda: identifying unsolved questions, designing hypotheses, and developing algorithms that could redefine what these models can do.
- Explore new architectures, learning objectives, and training paradigms that move beyond today’s diffusion and transformer approaches.
- Develop conceptual frameworks and experimental systems that deepen our understanding of how models perceive and respond to user intent and real-world dynamics.
- Collaborate closely with our engineering and infrastructure teams to turn foundational insights into prototypes that demonstrate new capabilities, not just incremental improvements.
- Contribute to the scientific culture of the lab: publishing, mentoring, and engaging with the broader research community to shape the field of interactive generative models.
Other
- deeply technical and creative researcher who thrives on invention
- motivated by the unknown, energized by turning abstract ideas into concrete systems, and comfortable bridging the gap between research and product.
- Thrives in a small, focused lab environment that values autonomy, deep thinking, and tight research collaboration.
- Passionate about establishing the scientific underpinnings of a new medium—helping define how AI perceives, learns, and interacts through pixels over the next decade.
