eBay is looking to enhance and optimize the marketplace experience by pushing the boundaries of artificial intelligence, and the Sr. Applied Research technical lead will be at the center of how multi-modal intelligence is applied across products.
Requirements
- Ph.D. in AI or equivalent research experience
- Proficiency in Python and PyTorch for data processing, model training and inference
- Expertise in Vision-Language Models and image/video retrieval
- Deep understanding of key computer vision techniques for classification, image/video retrieval, 3D modeling, and Image/Video generation
- Experience with frameworks for large-scale multi-GPU training (e.g. PyTorch Lightening, Nemo etc.)
- Experience with frameworks for large scale inference (TensorRT, vLLM )
- Two years plus of experience as a tech lead or a manager of science/engineering teams
Responsibilities
- Build and lead a high-functioning team of applied researchers and machine learning engineers
- Own the creation, management, and evolution of multimodal solutions used across company’s AI initiatives
- Design structured processes for data acquisition, model training, and inference
- Design, develop and improve large scale multimodal retrieval systems
- Maintain high levels of model quality through both offline and online evaluations
- Partner with internal teams and external providers to ensure that the right AI solutions and features are available for eBay use cases
- Establish and maintain clear standards for how data is collected, and models are trained and evaluations
Other
- Comfortable coaching, and enabling team members to deliver high-quality work through clear direction, support, and collaboration
- Ability to clearly communicate with technical and non-technical teams, making complex issues easy to understand without oversimplifying
- Ability to break down and clarify ambiguous problems, identify what’s missing, and help teams move toward practical, well-defined solutions
- Detail oriented in how you work with data, models, documentation, and process
- Ability to catch inconsistencies early, ensuring standards are consistently met across workflows and deliverables
