Principal Data Scientist, Gen AI and Vision

What's the role?We are hiring a Principal Data Scientist to lead advanced generative AI and computer vision work across simulation-grounded visual systems, structured-control generation, and production-oriented AI products. This person will help define and build model capabilities that sit at the intersection of generative modeling, scene understanding, controllability, and applied computer vision. This is a senior technical role for someone who can move from early research ambiguity to production-quality model systems while maintaining a high bar for technical depth, rigor, and practical value. This role is intended for someone who can own a major technical pillar, set direction for that domain, and raise the standard for how advanced AI systems are built and shipped. What You Will Own- Own the model strategy and technical direction for advanced generative and multimodal vision systems- Set the research and engineering agenda for model adaptation, evaluation, and production readiness within this domain- Define how model inputs, controls, outputs, and interfaces are represented and integrated into downstream systems- Improve realism, controllability, temporal consistency, robustness, and deployment readiness over time- Work closely with evaluation, simulation, and platform engineers to define durable interfaces between research systems and production workflows- Drive model benchmarking, ablation studies, model-selection decisions, and the path from pre-trained baselines to production-ready capability- Raise the bar for technical quality, architecture, and execution within the model stack What You Will Do - Build and evolve end-to-end model workflows for complex vision and generative AI systems- Identify and reduce failure modes such as structural drift, temporal instability, inconsistent outputs, and production fragility- Define the fine-tuning, adaptation, or hybrid-model strategy when pre-trained models do not meet product requirements out of the box- Partner with the evaluation lead to establish quality metrics, release gates, and production-readiness criteria- Help shape the roadmap from initial prototypes to scalable AI capabilities across perception, generation, and related vision tasks- Contribute to deployment decisions around model packaging, inference optimization, and production performance tradeoffs- Mentor other scientists and act as the senior technical owner for the model domain Who are you?What We Are Looking For:- Deep experience in generative modeling for video, world models, diffusion models, multimodal systems, or adjacent advanced vision domains- Evidence of principal-level scope through technical leadership, research impact, architectural ownership, or shipped systems- Strong background in PyTorch and modern model training, fine-tuning, evaluation, and inference workflows- Experience adapting large pre-trained models to domain-specific use cases and hard production constraints- Good judgment on model quality, controllability, reliability, and deployment tradeoffs- Ability to work across research and engineering boundaries in a small, hands-on team
Education & Experience

Skills Required
Deep experience in generative modeling for video, world models, diffusion models, or multimodal vision
Evidence of principal-level scope through technical leadership, research impact, architectural ownership, or shipped systems
Strong background in PyTorch and modern model training, fine-tuning, evaluation, and inference workflows
Experience adapting large pre-trained models to domain-specific use cases and production constraints
Good judgment on model quality, controllability, reliability, and deployment tradeoffs
Ability to work across research and engineering boundaries in a small, hands-on team and mentor others
Master's or PhD in Computer Science, AI, Machine Learning, or related field
10+ years of experience in deep learning, computer vision, or multimodal AI
Experience with synthetic data, robotics, perception systems, geospatial AI, autonomous systems, or mapping products
Familiarity with conditioning mechanisms, structured control inputs, simulation-grounded models, or controllable generation
Experience running large-model inference, optimization, or fine-tuning on AWS GPU infrastructure
Experience taking research models into production environments