Manli Shu

Visionary
DISC Type : Ds

Research Scientist at Google DeepMind

Palo Alto, California, United States

Overview

Manli Shu is a Research Scientist at Google DeepMind focusing on multimodal LLMs, AI safety, and trustworthiness. A Ph. D. graduate from the University of Maryland, she has interned at NVIDIA and Google and was a researcher at Salesforce, contributing to projects like BLIP-3 and presenting extensively at major AI conferences.

Manli is deeply engaged with the AI research community, frequently presenting her work at top conferences like NeurIPS. She is passionate about discussing complex topics in her field, from vision-language models to the broader challenges and opportunities within a Ph. D. program, and is open to new collaborations.

Her research includes developing novel ways to understand model vulnerabilities, such as creating stealthy data poisoning attacks on vision-language models.

Personality Overview

Objective Evaluator

Big Vision Person

Direct & Assertive

They are very professional in their approach and can weigh multiple perspectives together.  They might take some time to make their mind up but once they do, they don't change it easily. Reading between the lines and seeing beyond your words comes naturally to them.

Topics They Care About

Multimodal LLMs
Her core focus at Google DeepMind is on Gemini's multimodal understanding, and she previously led post-training efforts for Salesforce's BLIP-3 vision-language model.
AI/ML Trustworthiness
A primary theme of her Ph. D. research and ongoing work, focusing on AI safety in both vision and language modalities to create more reliable models.
LLM Security
She actively researches vulnerabilities in large language models, having published and presented papers on data poisoning attacks and adversarial attacks on LLMs at NeurIPS.

Media Appearances

Manli has no verified media appearances

Work History

6-2025
Research Scientist at Google DeepMind
1-2024 - 6-2025
Applied Scientist, AI Research at Salesforce
6-2023 - 11-2023
Research Intern at Google
6-2022 - 11-2022
Research Intern at Salesforce
1-2022 - 5-2022
Research Intern at NVIDIA

Education

2019 - 2023
Doctor of Philosophy - PhD from University of Maryland
2015 - 2019
Bachelor's degree from University of Science and Technology of China

More Information

Social Presence :

Prographics :

Exp : 2 Location : Palo Alto, California, United States Job Level : N/A Designation : Research Scientist at Google DeepMind
URL has been copied!

Insights For Selling To Manli

During A Call Or A Meeting

DO's

  • Use phrases like 'your team deserves', 'best in class' etc.
  • Focus on the results that your product produces, expect some strategic questions in return
  • Let them know of potential risks but suggest mitigation methods alongside

DONT's

  • Don't go over them unless you are left with no other option
  • Don't get into pricing discussions early on, steer conversation towards proven results
  • Avoid putting conscious effort into relationship-building

When Cold Calling

When Writing An Email

While Negotiating & Closing

    The secret to closing fast with Manli is

  • Strong proof of impact and their conviction will matter the most, but they wouldn't want to act unilaterally either
  • Will you ever get a clear answer from Manli

  • They will say no if they are not convinced but you will have to prompt them.

Insights For Deal Planning

    How fast (or slow) will Manli move?

  • They will want to understand things well but can move fast once they have a clear picture.
  • Can Manli take some risk or not?

  • They have good risk tolerance but are likely to think it through once or twice.

You And Manli

Personality Compatibility


Other Google DeepMind Employees

Explore more public profiles from related professionals at the same organization.

More Profiles

Discover additional public profiles from our index.

Search more profiles

Looking for someone else? Search here for anyone.

Or visit Humantic AI to know more.