Microsoft2025–present
LLM agents & generation
Developing LLM agents with memory, reasoning, and reflection, alongside presentation generation and evaluation.
Applied scientist · Mathematical foundations
Jinghua Yao
Principal Applied Scientist at Microsoft · Ph.D. in Mathematics
From mathematical rigor
to real-world AI.
I build AI systems grounded in mathematics—from LLM agents and multimodal learning to enterprise applications.
SAN JOSE, CALIFORNIA
A SCIENTIST’S PERSPECTIVE01 / 03
Rigor in the thinking.
Clarity in the making.RESEARCH × ENGINEERING
Clarity in the making.RESEARCH × ENGINEERING
Experience &
educationMicrosoftAppleSAMSUNGIndiana University
educationMicrosoftAppleSAMSUNGIndiana University
01 / Selected work
Ideas, built into systems.
Applied AI across Microsoft, Apple, and Samsung.
Apple2024
Multimodal understanding
Architected visual understanding with late interaction; developed distributed LLM/VLM fine-tuning for embedding and generative AI applications.
Apple2020–2023
Enterprise ML & automation
Built forecasting, anomaly detection, and transformer-based sentiment systems, alongside automated reporting, document Q&A, and summarization.
Samsung Display2019–2020
Computer vision for quality
Led a product-used display-defect classification solution, combining convolutional networks and classical ML with approximately 800 microscopy images.
02 / Experience
Research depth.
Industry perspective.
JUL 2025 — PRESENT
Microsoft
Principal Applied Scientist
LLM agent workflows, presentation generation, and evaluation for AI applications.
MAY 2020 — JUN 2025
Apple
Senior Machine Learning Engineer
Multimodal learning, distributed fine-tuning, enterprise LLM applications, forecasting, anomaly detection, and technical education.
MAR 2019 — MAY 2020
Samsung Display
Staff Machine Learning Engineer
Applied computer vision and machine learning for display-defect detection and classification.
2013 — 2017
Academic research
Nonlinear analysis & PDEs
Visiting Assistant Professor, University of Iowa (Aug 2013–Jan 2016); Research Scholar, Kent State University (Sep 2016–Sep 2017).
03 / Research
A foundation in mathematics.
Nonlinear analysis, partial differential equations, dynamical systems, and variational methods.
2019
Patterns of complex oscillations and instability in chemical reactions
Communications in Mathematical Sciences · 17(6), 1713–1736
2017
Spectrum and bifurcation for semilinear elliptic problems in ℝᴺ
Journal of Differential Equations · 263(9), 5939–5967
2014
O(2)-Hopf bifurcation for a model of cellular shock instability
Physica D: Nonlinear Phenomena · 269, 63–75
04 / About
A mathematician
who builds.
I earned my Ph.D. in Mathematics at Indiana University Bloomington and worked in academic research before moving into industrial machine learning.
I connect rigorous problem formulation with practical engineering—making assumptions explicit and complex ideas clear.
Ph.D. in MathematicsIndiana University Bloomington
Master’s in MathematicsLanzhou University
Bachelor’s in MathematicsLanzhou University
Technical range
Generative AI
LLM agents · LLM/VLM fine-tuning · Transformers · Multimodal understanding · Embeddings · Document Q&A · Summarization
Applied machine learning
Forecasting · Anomaly detection · NLP · Computer vision · Statistical modeling · Optimization
Engineering
Python · PySpark · C# · TypeScript · Go · scikit-learn · Docker · AWS · Git · Streamlit · Gradio
Research community
Invited reviewer for Mathematical Reviews; journal referee in mathematical analysis and related fields.
Let’s connect
Good problems deserve
thoughtful solutions.
For applied science opportunities and research collaboration, I’d be glad to connect.
jinghuayao AT gmail DOT comFind me on GitHub ↗