AI Research Scientist — Generative Vision & Virtual Try-On
About StyonStyon is an AI stylist platform building AI-powered outfit recommendations, virtual try-on, and virtual wardrobes. We are developing models that help people visualize how clothing looks on their own bodies, with a focus on realistic appearance, garment fidelity, and a personalized experience. Role OverviewWe are looking for a PhD researcher with a strong background in computer vision to advance our virtual try-on research on a contract or internship basis. Prior VTO project or research experience is highly preferred. This is a research-first role with a clear path to product impact. You will explore challenging VTO problems, prototype new approaches, and help translate promising research into reliable experiences for users.
What You Will Do
Research and prototype virtual try-on methods, particularly diffusion-based approaches, with a focus on garment-person alignment, garment fidelity, and identity and pose preservation. Improve the preservation of garment textures, patterns, logos, and structural details. Address challenging try-on scenarios, including occlusions, complex poses, layered outfits, and diverse body shapes and skin tones. Explore garment deformation, human representations, and 3D-aware approaches where relevant to improving try-on quality. Design evaluations for visual realism, garment fidelity, identity preservation, and consistency across poses and views. Reproduce and extend state-of-the-art methods, run controlled experiments, and clearly document results, limitations, and trade-offs. Collaborate with Styon’s AI and product teams to move promising prototypes toward production, balancing quality, latency, and cost. Required QualificationsCurrently enrolled in, or recently graduated from, a PhD program in computer science, electrical engineering, machine learning, or a related field. Strong research background and hands-on project experience in computer vision. Strong Python and PyTorch skills, with experience implementing, training, and evaluating deep learning models. Research or project experience in at least one relevant area, such as image generation or editing, diffusion models, human-centric vision, image synthesis, or 3D vision. Ability to translate open-ended research problems into testable hypotheses, well-designed experiments, and clear conclusions. Strong communication skills and the ability to work independently in a remote, early-stage team. Preferred QualificationsPrior virtual try-on research or project experience, ideally involving hands-on model development and evaluation. Experience with garment warping, human parsing, DensePose, pose transfer, human generation, or garment modeling. Experience with VTO datasets, 3D human or garment reconstruction, or body and cloth representations. Publications at CVPR, ICCV, ECCV, NeurIPS, ICLR, or comparable venues in relevant research areas. Experience with inference optimization or deploying image generation models. Open-source contributions, research code, or demos demonstrating VTO or related computer vision work. Contract DetailsLocation: United States, remote Type: Research contract or internship, suitable for a current PhD student or recent PhD graduate Duration: 3 months, with the possibility of extension or conversion to a permanent, full-time role based on performance, mutual fit, and business needs. Schedule: Flexible, with hours agreed upon based on project needs and academic commitments Compensation: Competitive, based on experience and engagement scope