Lead Generative AI Engineer (3D, VLM & Diffusion Models)

Job Description

Company Description

Edensign is building the future of AI-powered visual and spatial design. Backed by the Harvard Innovation Labs, we’re creating next-generation intelligent systems that merge generative AI, 3D understanding, and spatial intelligence to transform how real-world spaces are visualized, staged, and experienced.

Contact Email: hello@edensign.io

Role Description

Full-time | Preference for Boston or New York–based candidates

We’re looking for a senior technical leader to drive the development of our core AI engine. The ideal candidate has deep experience training large generative models, including diffusion, 3D reconstruction networks, multimodal, VLM architectures. In this role, you will spearhead model training pipelines, R&D experiments, data strategy, and foundational architecture decisions.

This is an opportunity to help build the next generation of spatial AI - from multi-view consistency to 2D-to-3D-to-2D transformation and advanced scene understanding.

Key Responsibilities

• Design, train, and optimize cutting-edge generative models, including diffusion, 3D reconstruction, and multimodal/VLM architectures
• Build and manage scalable training pipelines, data curation workflows, and experiment tracking
• Lead research experiments, benchmarking, and exploration of new modeling techniques
• Architect the evolution of our spatial AI stack—from prototyping new ideas to deploying production-ready models
• Collaborate with engineering and product teams to integrate AI capabilities seamlessly into real-world workflows
• Make strategic decisions around infrastructure, GPU utilization, model efficiency, and training optimization
• Contribute to Edensign’s long-term technical roadmap and innovation direction
  

Qualifications

• Strong expertise in training generative models (diffusion, GANs, 3D generative models, or scene-reconstruction networks)
• Deep background in Computer Vision, 3D geometry, NeRF-like architectures, or multi-view learning
• Experience with VLMs, multimodal models, grounding, or spatial reasoning is highly valuable
• Proficiency in Python and modern ML frameworks (PyTorch preferred)
• Hands-on experience with distributed training, GPU optimization, and large-scale experiment management
• Familiarity with node-based generative tools (e.g., ComfyUI) is a plus
• Ability to work independently and lead technical direction in a fast-paced startup environment
• Strong analytical, problem-solving, and system design skills
• Excellent communication and collaboration skills
• Master’s or PhD in Computer Science, AI/ML, Computer Vision, Robotics, or a related field
• Experience in real estate, architecture, spatial design, or spatial computing is a bonus