Research Scientist, Real-Time DiT Video Models

Remote; San Francisco (CA); On-site; USD 180,000 - 240,000; Full time

Job summary

Valka is recruiting a research scientist to lead the development of interactive video content using diffusion-based video synthesis. You will push the boundaries of realism, consistency, and style in generated video content and collaborate across teams to align research with product roadmap.

The role requires a PhD in computer vision or ML, strong publications, and hands-on experience with Python and PyTorch.

Qualifications

• PhD in Computer Vision, ML, or related field.
• Hands-on research on video diffusion models.
• Strong publication record at top conferences (CVPR/NeurIPS/ICCV).

Responsibilities

• Develop and optimize video diffusion models for realistic, controllable results.
• Advance state-of-the-art algorithms for video generation.
• Collaborate with teams on large-scale video datasets including motion and gestures.
• Experiment with diffusion architectures for high-quality synthesis.
• Create validation protocols and metrics comparing synthetic vs real gameplay.
• Stay current with CVPR, NeurIPS, ICML, ICCV literature and apply insights.

Skills

Python programming

PyTorch

Video diffusion models

Research literacy

Education

PhD in Computer Vision or ML

Tools

PyTorch

TensorFlow

CUDA

Job description

About Valka

Valka, a visionary spin-off from the Realms Group (the parent company of Oddin.gg), is on a mission to revolutionize the way people create and experience digital content. Our team believes that content shouldn’t just be consumed; it should be co-created in real time, blurring the lines between imagination and reality. By harnessing the power of cutting-edge AI, we aim to build an interactive human-digital platform where virtual characters respond dynamically to each user’s voice, text, gestures, and more.

This is your chance to join a diverse group of innovators who are driven to redefine what’s possible in generative content. Together, we’re changing the paradigm from passive viewing to active participation, unlocking new creative frontiers across gaming, entertainment, education, and beyond.

Key Responsibilities:

• Design, develop, and optimize AI video generation models using diffusion techniques, with a focus on maintaining consistency, realism, and style.

• Develop and implement state-of-the-art algorithms for generating video content.

• Work closely with other teams on large-scale video datasets, including human motion and gestures, facial expressions, and scene context.

• Experiment with cutting-edge diffusion architectures for controllable and high-quality video synthesis.

• Define robust validation strategies and implement custom evaluation metrics comparing synthetic vs. real gameplay.

• Stay on the bleeding edge of the relevant literature, e.g., CVPR, NeurIPS, ICML, ICCV, and help to align it with our roadmap.

Required Qualifications:

• PhD in Computer Vision, Machine Learning, or a closely related field.

• Hands-on experience with research of video diffusion models.

• Strong research background in image or video synthesis demonstrated by having publications at top tier conferences such as CVPR, NeurIPS, ICCV, Siggraph, and ICML.

• Experience with real time video generation is a big plus.

• Proficiency in Python and ML frameworks such as PyTorch

• Solid understanding of neural architectures or paradigms (CNNs, Transformers, diffusion models, autoregressive models etc.).

Join us at Valka to lead a new wave of interactive video content, one where your creativity and technical prowess will help transform entire industries and reimagine how digital content is created, shared, and experienced.

This role offers a unique opportunity to shape the future of interactive video content, where digital humans can engage in meaningful and dynamic interactions with users. If you're a passionate ML expert with a drive to innovate and create immersive experiences.