Adversarial robustness
Designing architectures and training strategies that improve resilience without making reliable AI prohibitively expensive.
Machine learning · robust systems · applied research
Korn
Sooksatra
I build machine learning systems that stay reliable under pressure.
A researcher and engineer working across adversarial robustness, recommendation systems, computer vision, NLP, speech, and reinforcement learning—with experience translating research into production-scale systems.
5+years in ML research
Ph.D.Computer Science · Baylor
15+authored and co-authored works
4industry research environments
01 / FOCUS
Making intelligent systems more dependable.
My work centers on the gap between benchmark performance and real-world reliability: how models behave under corrupted signals, hostile inputs, sparse feedback, and system constraints.
Ranking & recommendation
Building data and learning systems for ad allocation, whole-page optimization, and decision-making under changing conditions.
Multimodal intelligence
Applying machine learning to event streams, language, vision, and time-series data—from fraud detection to satellite imagery.
Learning for complex systems
Using deep reinforcement learning and graph-based methods to reason about scheduling, software architecture, and control.
02 / EXPERIENCE
From research prototypes to operating systems.
Industry work across ranking, fraud, computer vision, and supply-chain optimization—grounded in a research practice that asks how systems fail before they ship.
SEP 2024 — PRESENT
Research Scientist
MetaDeveloping and maintaining technologies that improve the robustness of ads ranking models, including feature-corruption simulation and resilient model evaluation.
MAY — AUG 2024
Ph.D. Research Intern
PinterestCreated features, revenue models, and reinforcement-learning approaches for whole-page advertisement allocation and optimization.
MAY — AUG 2023
Research Intern
UberBuilt multimodal Transformer approaches for spender-fraud detection using event and timestamp streams, supported by production-oriented data processing.
2022
Machine Learning Intern
MappointAsia · AspenTech
Developed satellite-imagery ship detection with YOLOv4 and deep-RL solutions for NP-complete product scheduling in supply-chain systems.
03 / PUBLICATIONS
Selected research.
Work spanning robust activation functions, adversarial examples, neural cryptography, reinforcement learning, and software architecture.
2024
Dynamic-Max-Value ReLU Functions for Adversarially Robust Machine Learning Models
2023
Is ReLU Adversarially Robust?
2022
On Adversarial Examples for Text Classification by Perturbing Latent Representations
2021
Enhancing Adversarial Examples on Deep Q Networks with Previous Information
2021
An Adversarial Neural Cryptography Approach to Integrity Checking
04 / BACKGROUND
Technical depth, broad application.
EDUCATION
Ph.D. in Computer ScienceBaylor University · 2024 · GPA 3.85
M.S. in Computer ScienceGeorgia State University · 2020 · GPA 3.83
B.Eng. in Computer EngineeringKasetsart University · 2014 · First-Class Honors
TOOLKIT
CorePython, PyTorch, TensorFlow, scikit-learn, SQL
ScalePresto, PySpark, Ray, Docker, Google Cloud Platform
MethodsTransformers, reinforcement learning, adversarial ML, computer vision, GNNs
Let’s build AI that earns trust.
I’m interested in ambitious research and engineering problems across robust machine learning, computer vision, NLP, speech, and recommendation systems.
PUBLIC RESEARCH SOURCES
Rivas AI Lab · Baylor University · Mathematics / MDPI