Wanzheng Zhu

Software Engineer @ Google DeepMind

About

I work on Gemini Spark and Deep Research at Google DeepMind. - Gemini Spark: https://gemini.google/overview/agent/spark- Deep Research: https://blog.google/innovation-and-ai/models-and-research/gemini-models/next-generation-gemini-deep-research, https://blog.google/products/search/deep-search-business-calling-google-search, https://blog.google/innovation-and-ai/technology/developers-tools/deep-research-agent-gemini-apiSkills: Agent, Evaluation, Deep Research, Trajectory & Response Generation, LLM Post-training, Reasoning, Factuality & Grounding, Routing, NLP

Country

United States

City

Mountain View

Industry

Research

Skill

Data Mining, Natural Language Processing (NLP), Machine Learning, Computer Vision, Text Mining

Experience

Software Engineer

Google DeepMind

2022-5 - Present · 4 yrs 4 mos

Mountain View, CA

- Gemini Spark: https://gemini.google/overview/agent/spark - Deep Research: https://blog.google/innovation-and-ai/models-and-research/gemini-models/next-generation-gemini-deep-research, https://blog.google/products/search/deep-search-business-calling-google-search, https://blog.google/innovation-and-ai/technology/developers-tools/deep-research-agent-gemini-api - Model Routing: Directing a user's query to the most appropriate LLM to serve. - Model Evaluation: Evaluating the quality (general helpfulness, trust & safety, factuality, grounding) of the LLM generated output. - Mobile Notification: Recommending the most interesting and engaging content to users.

Research Assistant

University of Illinois Urbana-Champaign

2016-8 - 2022-5 · 5 yrs 10 mos

Urbana-Champaign, Illinois Area

- Conversational AI: engaging in natural, human-like conversations (more diverse, relevant and grammatically correct) - Euphemism (Code Name) Detection and Understanding: Identifying and interpreting coded language to uncover its hidden meaning on Social Media. - Human Mobility Modeling: Understanding and predicting people's movement with multi-dimensional data (time, location, text) on a spherical space. - Evaluation Metric: evaluating linguistic quality of LLM generated text by measuring grammar, redundancy, focus, and coherence. - Knowledge Base Construction: Constructing a hierarchical knowledge graph on raw text corpus.