Jiang Yu, Ph.D.

Manager, Applied Research | Engineering Manager (ML)

ML Leadership at Scale: Engineering Manager specializing in building and leading cross-functional teams to ship high-impact ML systems. Expert in multi-stage search ranking, recommendation engines, and LLM-powered agentic workflows.

Currently leading Applied Research at eBay, I focus on turning ambiguous product needs into technical roadmaps that deliver reliable, low-latency performance at scale.

$1.3B+

Incremental Growth Driven

20x

Training Iteration Velocity

80%+

Execution Completion Rate

Professional Leadership

eBay — Manager, Applied Research (Search & AI) Jun 2023 – Present
  • Team Leadership: Lead a cross-functional team advancing ML and information retrieval for core search ranking systems across retrieval, ranking, and reranking.
  • Business Impact: Contributed to surpassing $1B 2024 revenue growth targets and delivered 86% of 2025 revenue goals in H1.
  • Technical Innovation: Led Mixture-of-Ranking-Experts and RL-based reranking generating $300M+ in incremental annual revenue.
  • Agentic AI: Set the direction for Agentic AI in Search Ranking to enable context-aware, journey-optimized experiences.
  • Recognition: Recipient of eBay's inaugural AI Business Impact Award (2025), presented by the CEO.
Amazon — Applied Scientist (Lead), Alexa AI Jun 2020 – Jun 2023
  • Engineering Management: Led 7+ scientists and engineers building multilingual conversational systems across ASR, MT, and NLU.
  • System Quality: Improved Alexa Translation by reducing false rejections by 10% and repeat failures by 15%.
  • Product Launch: Owned Conversational Language Learning from concept through funding, prototype, and beta launch.
  • Research Excellence: Won 1st place at IWSLT 2022 for formality-control MT research.
Additional Industry Experience
  • Twitter: Improved Recommendation Core Engagement (RCE) by 33% and launched personalization features.
  • Tencent: Built a healthcare chatbot deployed to WeChat apps using RAG, LSTM, GRU and Transformers.
  • MSKCC: Developed deep learning models for clinical sequence prediction and clinical research database based web application.
  • Intel: Optimized distributed training over large-scale GPUs & CPUs for Transformer-based Lanague Models and Mixture-of-Experts models.
  • G-wear: Founded and led an AI-driven IoT safety monitoring startup (fall/dehydration detection) that won 1st place at the NYU Healthcare Makerthon and received seed funding from the Entrepreneur Lab Incubation.
  • Amazon AWS: Developed global call center service cloud integration, later launched as AWS Connect.

Selected Publications & Patents

Contextual Mixture of Ranker models Submitted to SIGIR 2026
US Patent IP-P3928PRV: Enhancing Buyer Engagement on Related Query Recommendation via LLM U.S. Patent Application

Education & Professional Development

New York University Ph.D. Computer Science
Stanford University GSB Executive Education