微软Senior Applied Scientist(Feeds and AI team)

• Drives product impact by transferring cutting-edge research into production systems. Collaborates across Microsoft and external groups to apply scalable, data-driven solutions, author white papers, file patents, and consult on product strategy.  • Designs and optimizes ranking algorithms across L1–L3 stages using collaborative filtering, deep neural networks (DNN), and graph neural networks (GNN). Applies FastRank and LambdaMart techniques for precision ranking in Bing Feeds and other verticals.  • Leads experimentation and deployment of large-scale NLP and LLM models for topic modeling, sentiment analysis, and multimodal content linking. Supports real-time indexing and modular APIs for rapid model onboarding.  • Cleans, transforms, and analyzes large-scale data to build usable datasets and optimize feature selection. Develops ETL pipelines and mentors teams in data preparation and signal system design.  • Operates on distributed GPU/CPU frameworks (DLTS, DeepSpeed) for training and inference. Supports A/B testing, flighting, and real-time feedback loops to refine models and improve user engagement.  • Applies deep understanding of fairness and bias in model development. Contributes to internal ethics policies and ensures representative data usage aligned with responsible AI principles.  • Coaches junior team members, collaborates with academia to recruit top talent, and shares research findings through publications and conferences. Builds multidisciplinary teams and fosters innovation culture. 

• Ship features with PM & Engineering. Co‑own scenario goals; translate product requirements into scientific plans and productionized solutions that meet quality/latency/cost targets. • Model development & optimization. Design, fine‑tune, and evaluate models for LLM‑based authoring, summarization, reasoning, voice/chat, and personalization (e.g., SFT, alignment, prompt/tool use, safety filtering, multilingual & multimodal). • Data & evaluation at scale. Build/extend data pipelines for curation/labeling/feature stores; author offline eval harnesses; run online A/Bs and interleavings; define guardrails and success metrics; author scorecards and decision memos. • Production ML engineering. contribute to service code and configs; add monitoring, tracing, dashboards, and auto‑scaling; participate in on‑call and postmortems to improve live‑site reliability. • Responsible AI. Produce review artifacts, document mitigations for safety/privacy/fairness, support red‑teaming and sensitive‑use checks, and align with Microsoft’s Responsible AI Standard. • Collaboration & mentoring. Partner across PM/ENG/Design/CE/ORA/CELA; share methods and code, review PRs, improve reproducibility and documentation; mentor junior scientists.

• Apply state-of-the-art research and advanced algorithms to develop scalable, data-driven solutions that deliver measurable product impact. • Collaborate with product, engineering, and research teams to transfer technology and integrate innovative approaches into production systems. • Design and optimize machine learning models and data pipelines for large-scale applications, ensuring performance, scalability, and ethical standards. • Mentor and guide less experienced team members, fostering technical growth and sharing best practices across projects. • Stay current with industry trends and emerging technologies; publish research and share insights to advance innovation and business impact. • Incorporate fairness, bias detection, and privacy considerations into research and product development processes. • Drive improvements in data quality and leverage advanced analytics to identify opportunities for product enhancement.

• Design and implement offline evaluation strategies that capture real-world usage and reflect end-user preferences. • Develop scientifically sound metrics that diagnose model regressions, benchmark against baselines (e.g., ChatGPT, Glean), and validate product improvements. • Manufacture synthetic yet realistic user activity data using LLMs to simulate diverse usage scenarios. • Collaborate on multi-agent systems or agentic workflows to automate evaluation flows and generate high-signal insights. • Analyze evaluation outputs to identify gaps in coverage, quality, and usability across Copilot canvases. • Partner with engineering and PMs to ensure insights are integrated into product workflows and experimentation pipelines. • Publish learnings in internal forums, external conferences, and contribute to best practices in applied science.