Onyx
ResearchTeam

Research

Onyx focuses heavily on AI research. Our security technology, Equivariant Encryption, is the result of countless hours of investigation and implementation in state-of-the-art privacy preserving technologies today.

Some of our recent research:

Towards Secure and Private AI: A Framework for AI Inference, Accepted to NeurIPS 2024, RBFM
Meta-Learning for Speeding Up Large Model Inference, Accepted to AAAI 2025, DAI
Model Agnostic Hybrid Sharding for Heterogeneous Inference, Accepted to KDD 2024, SDBD
200+

Published research papers

70+

Conference acceptances

8

1st place finishes

25+

ML-TCS Area Chairs

Team Onyx

Research Awards

Alan Turing
NeurIPS
NeurIPS
CVPR
Google
Meta
Google
ICML
IEEE
Area Chair
ICLR
Norton
Github
WiSE Merit
ACM CCS
Embassy
Baidu
Nature Science
MIT EECS
AAAI
MIT-Takeda
Wold AI WAICO

Research Team

Dr. Harry Yang

Dr. Harry Yang

Founder, CTO

Facebook AI. Creator of Facebook's Sora, called Make-a-Video. Early LLM development work at Meta for Llama. Computer vision and video gen AI expert. Award winning researcher. PhD in AI and MSc in CS, USC.

Dr. Yue Zhao

Dr. Yue Zhao

Chief Scientist

Professor of AI, USC. Creator of PyOD (22 million downloads) the most famous anomaly detection AI used by Tesla, NASA. 20k+ Github stars (top 0.001%). AAAI New Faculty Highlight. Google and Meta AI Research Awards. PhD in AI, Carnegie Mellon.

Dr. Claudio Angione

Dr. Claudio Angione

Chief of Research

Alan Turing Development Award recipient in AI. PhD in CS, University of Cambridge. MSc and BSc in Mathematics. Postdoc AI research, University of Cambridge. Former Microsoft Research Scientist. Professor of AI, Teesside University.

Dr. Marco Di Maggio

Dr. Marco Di Maggio

Founder, CSO

Director, Harvard Crypto & Web3 Lab. Former Professor, Harvard Business School. Professor of Finance, Imperial College London. Fellow, National Bureau of Economic Research. PhD in Economics, MIT.

Patrick Colangelo

Patrick Colangelo

Founder, CEO

Research in NLP. Co-Creator of the first transparent cubesat for spaceflight, permanently installed on the Moon. Work with LUNR and NASA. Harvard grad.

Dr. Yuzhe Yang

Dr. Yuzhe Yang

AI Chief

Forbes 30 under 30. Phd in CS, MIT. Named Rising Star in AI. Research honored to World's 10 Notable Advances, Nature Science. Trustworthy ML expert. BSc in CS, Peking. Baidu PhD scholar. NeurIPS Top Reviewer.

Dr. Hongyang Zhang

Dr. Hongyang Zhang

Chief of Security

One of the fathers of ZKML. Creator of ZKDL. Creator of zkLLM. Creator of EAGLE, the fastest LLM decoder in the world. Creator of RAIN. 1st Place at NeurIPS, 1st Place at CVPR. Professor of AI, UWaterloo.

Dr. James Buban

Dr. James Buban

Systems Lead

Earliest Meta faster R-CNN engineer. Research Professor of AI, University of Illinois. PhD in Physics, MSc in Mathematics, BSc in Physics. 10 year AI Research Scientist.

Fielding Johnston

Fielding Johnston

VP of Engineering

Ex VP eng. 15 years of experience in engineering, blockchain, and machine learning. Langchain framework contributor. Billions of transactions of scale.

Ahmad Farhan

Ahmad Farhan

AI Lead

Postponed a PhD in AI to lead Nesa. MSc in AI. BSc in CS. Pakistan's 1 out of 100 National Scholarship Winner. Provincial Benevolent Fund Scholar. Half decade in deep learning.

Dr. Jennifer Wang

Dr. Jennifer Wang

Security Lead

Named MIT EECS Rising Star. ZK expert. Professor of AI, Purdue. Postdoc research in AI Security and ZK. PhD in CS, USC. ACM Distinguished Paper Award. Winner, WiSE Merit Award.

Dr. Hao Li

Dr. Hao Li

Research Science

Co-Creator of Animoji on Apple’s iPhone X. Named one of the top 35 innovators under 35 by MIT Tech Review. Award winning AI researcher. Founder and CEO of Pinscreen. Professor of AI, MBZUAI and USC.

Harward University
Meta AI
MIT
USC
Imperial
Facebook
UC
Microsoft
TU
CMU
The Alan Turing Institute
University of Waterloo
ByteDance
The Hong Kong University of Science and Technology
Nber

Research by the Team

Make-a-video: Text-to-video generation without text-video data

Make-a-video: Text-to-video generation without text-video data

Dr. Harry Yang

Diffusion Models: A Comprehensive Survey of Methods and Applications

Diffusion Models: A Comprehensive Survey of Methods and Applications

Dr. Yue Zhao

zkLLM: Zero Knowledge Proofs for Large Language Models

zkLLM: Zero Knowledge Proofs for Large Language Models

Dr. Hongyang Zhang

PyOD: A Python Toolbox for Scalable Outlier Detection

PyOD: A Python Toolbox for Scalable Outlier Detection

Dr. Yue Zhao

Contextual-based image inpainting: Infer, match, and translate

Contextual-based image inpainting: Infer, match, and translate

Dr. Harry Yang

zkDL: Efficient Zero-Knowledge Proofs of Deep Learning Training

zkDL: Efficient Zero-Knowledge Proofs of Deep Learning Training

Dr. Hongyang Zhang

Long video generation with time-agnostic vqgan and time-sensitive transformer

Long video generation with time-agnostic vqgan and time-sensitive transformer

Dr. Harry Yang

Theoretically Principled Trade-off between Robustness and Accuracy

Theoretically Principled Trade-off between Robustness and Accuracy

Dr. Hongyang Zhang

Using machine learning as a surrogate model for agent-based simulations

Using machine learning as a surrogate model for agent-based simulations

Dr. Claudio Angione

A Closer Look at Accuracy vs. Robustness

A Closer Look at Accuracy vs. Robustness

Dr. Hongyang Zhang

Rethinking the Value of Labels for Improving Class-Imbalanced Learning

Rethinking the Value of Labels for Improving Class-Imbalanced Learning

Dr. Yuzhe Yang

COPOD: Copula-Based Outlier Detection

COPOD: Copula-Based Outlier Detection

Dr. Yue Zhao

ReluDiff: differential verification of deep neural networks

ReluDiff: differential verification of deep neural networks

Dr. Jingbo Wang

On the Applications of Robust PCA in Image and Video Processing

On the Applications of Robust PCA in Image and Video Processing

Dr. Hongyang Zhang

ADBench: Anomaly Detection Benchmark

ADBench: Anomaly Detection Benchmark

Dr. Yue Zhao

High-resolution image inpainting using multi-scale neural patch synthesis

High-resolution image inpainting using multi-scale neural patch synthesis

Dr. Harry Yang

A pipeline and comparative study of 12 machine learning models for text classification

A pipeline and comparative study of 12 machine learning models for text classification

Dr. Claudio Angione

Self-Adaptive Training: beyond Empirical Risk Minimization

Self-Adaptive Training: beyond Empirical Risk Minimization

Dr. Hongyang Zhang

Proceedings of Machine Learning Research

Proceedings of Machine Learning Research

Dr. Yuzhe Yang

ECOD: Unsupervised Outlier Detection Using Empirical Cumulative Distribution Functions

ECOD: Unsupervised Outlier Detection Using Empirical Cumulative Distribution Functions

Dr. Yue Zhao

NeuroDiff: scalable differential verification of neural networks using fine-grained approximation

NeuroDiff: scalable differential verification of neural networks using fine-grained approximation

Dr. Jingbo Wang

Learning and 1-bit Compressed Sensing under Asymmetric Noise

Learning and 1-bit Compressed Sensing under Asymmetric Noise

Dr. Hongyang Zhang

Spg-net: Segmentation prediction and guidance network for image inpainting

Spg-net: Segmentation prediction and guidance network for image inpainting

Dr. Harry YangDr. Yuhang Song

Therapeutics Data Commons: Machine Learning Datasets and Tasks for Drug Discovery and Development

Therapeutics Data Commons: Machine Learning Datasets and Tasks for Drug Discovery and Development

Dr. Yue Zhao

ME-Net: Towards Effective Adversarial Robustness with Matrix Estimation

ME-Net: Towards Effective Adversarial Robustness with Matrix Estimation

Dr. Yuzhe Yang

Differentially Private Clustering in High-Dimensional Euclidean Spaces

Differentially Private Clustering in High-Dimensional Euclidean Spaces

Dr. Hongyang Zhang

Latent-shift: Latent diffusion with temporal shift for efficient text-to-video generation

Latent-shift: Latent diffusion with temporal shift for efficient text-to-video generation

Dr. Harry Yang

Mitigating power side channels during compilation

Mitigating power side channels during compilation

Dr. Jingbo Wang

Employee Turnover Prediction with Machine Learning: A Reliable Approach

Employee Turnover Prediction with Machine Learning: A Reliable Approach

Dr. Yue Zhao

ME-Net: Towards Effective Adversarial Robustness with Matrix Estimation

ME-Net: Towards Effective Adversarial Robustness with Matrix Estimation

Dr. Yuzhe Yang

Shield Synthesis for Real: Enforcing Safety in Cyber-Physical Systems

Shield Synthesis for Real: Enforcing Safety in Cyber-Physical Systems

Dr. Jingbo Wang

Random Smoothing Might be Unable to Certify ℓ∞ Robustness for High-Dimensional Images

Random Smoothing Might be Unable to Certify ℓ∞ Robustness for High-Dimensional Images

Dr. Hongyang Zhang

XGBOD: Improving Supervised Outlier Detection with Unsupervised Representation Learning

XGBOD: Improving Supervised Outlier Detection with Unsupervised Representation Learning

Dr. Yue Zhao

Shape inpainting using 3d generative adversarial network and recurrent convolutional networks

Shape inpainting using 3d generative adversarial network and recurrent convolutional networks

Dr. Harry Yang

Targeted Supervised Contrastive Learning for Long-Tailed Recognition

Targeted Supervised Contrastive Learning for Long-Tailed Recognition

Dr. Yuzhe Yang

LSCP: Locally selective combination in parallel outlier ensembles

LSCP: Locally selective combination in parallel outlier ensembles

Dr. Yue Zhao

Robust latent low rank representation for subspace clustering

Robust latent low rank representation for subspace clustering

Dr. Hongyang Zhang

When NAS Meets Robustness: In Search of Robust Architectures Against Adversarial Attacks

When NAS Meets Robustness: In Search of Robust Architectures Against Adversarial Attacks

Dr. Yuzhe Yang

Revisiting Time Series Outlier Detection: Definitions and Benchmarks

Revisiting Time Series Outlier Detection: Definitions and Benchmarks

Dr. Yue Zhao

Adversarial Vision Challenge

Adversarial Vision Challenge

Dr. Hongyang Zhang

Synthesizing Fair Decision Trees via Iterative Constraint Solving

Synthesizing Fair Decision Trees via Iterative Constraint Solving

Dr. Jingbo Wang

Real-Time Profiling of Fine-Grained Air Quality Index Distribution Using UAV Sensing

Real-Time Profiling of Fine-Grained Air Quality Index Distribution Using UAV Sensing

Dr. Yuzhe Yang

Rain: Your language models can align themselves without finetuning

Rain: Your language models can align themselves without finetuning

Dr. Hongyang Zhang

Realistic dynamic facial textures from a single image using gans

Realistic dynamic facial textures from a single image using gans

Dr. Harry Yang

TrustLLM: Trustworthiness in Large Language Models

TrustLLM: Trustworthiness in Large Language Models

Dr. Yue Zhao

UAV Aided Aerial-Ground IoT for Air Quality Sensing in Smart City: Architecture, Technologies, and Implementation

UAV Aided Aerial-Ground IoT for Air Quality Sensing in Smart City: Architecture, Technologies, and Implementation

Dr. Yuzhe Yang

Exact Recoverability of Robust PCA via Outlier Pursuit with Tight Recovery Bounds

Exact Recoverability of Robust PCA via Outlier Pursuit with Tight Recovery Bounds

Dr. Hongyang Zhang

Data-Driven Synthesis of Provably Sound Side Channel Analyses

Data-Driven Synthesis of Provably Sound Side Channel Analyses

Dr. Jingbo Wang

Automatic Unsupervised Outlier Model Selection

Automatic Unsupervised Outlier Model Selection

Dr. Yue Zhao

BodyCompass: Monitoring Sleep Posture with Wireless Signals

BodyCompass: Monitoring Sleep Posture with Wireless Signals

Dr. Yuzhe Yang

Understanding deep representations through random weights

Understanding deep representations through random weights

Dr. Jingbo Wang

Ordinal regression with neuron stick-breaking for medical diagnosis

Ordinal regression with neuron stick-breaking for medical diagnosis

Dr. Harry Yang

Noise-Tolerant Life-Long Matrix Completion via Adaptive Sampling

Noise-Tolerant Life-Long Matrix Completion via Adaptive Sampling

Dr. Hongyang Zhang

TODS: An Automated Time Series Outlier Detection System

TODS: An Automated Time Series Outlier Detection System

Dr. Yue Zhao

Change is Hard: A Closer Look at Subpopulation Shift

Change is Hard: A Closer Look at Subpopulation Shift

Dr. Yuzhe Yang

Deep Neural Networks with Multi-Branch Architectures Are Intrinsically Less Non-Convex

Deep Neural Networks with Multi-Branch Architectures Are Intrinsically Less Non-Convex

Dr. Hongyang Zhang

Attacks and design of image recognition captchas

Attacks and design of image recognition captchas

Dr. Harry Yang

Systematic Testing of the Data-Poisoning Robustness of KNN

Systematic Testing of the Data-Poisoning Robustness of KNN

Dr. Jingbo Wang

On Multi-Domain Long-Tailed Recognition, Imbalanced Domain Generalization and Beyond

On Multi-Domain Long-Tailed Recognition, Imbalanced Domain Generalization and Beyond

Dr. Yuzhe Yang

SUOD: Accelerating Large-Scale Unsupervised Heterogeneous Outlier Detection

SUOD: Accelerating Large-Scale Unsupervised Heterogeneous Outlier Detection

Dr. Yue Zhao

Noise-Tolerant Interactive Learning Using Pairwise Comparisons

Noise-Tolerant Interactive Learning Using Pairwise Comparisons

Dr. Hongyang Zhang

Monitoring gait at home with radio waves in Parkinson’s disease: A marker of severity, progression, and medication response

Monitoring gait at home with radio waves in Parkinson’s disease: A marker of severity, progression, and medication response

Dr. Yuzhe Yang

BOND: Benchmarking Unsupervised Outlier Node Detection on Static Attributed Graphs

BOND: Benchmarking Unsupervised Outlier Node Detection on Static Attributed Graphs

Dr. Yue Zhao

Sample and Computationally Efficient Learning Algorithms under S-Concave Distributions

Sample and Computationally Efficient Learning Algorithms under S-Concave Distributions

Dr. Hongyang Zhang

Regmixup: Mixup as a regularizer can surprisingly improve accuracy and out distribution robustness

Regmixup: Mixup as a regularizer can surprisingly improve accuracy and out distribution robustness

Dr. Harry Yang

Contrastive Attributed Network Anomaly Detection with Data Augmentation

Contrastive Attributed Network Anomaly Detection with Data Augmentation

Dr. Yue Zhao

ResearchTeam
COPYRIGHT © 2025 ONYX. ALL RIGHTS RESERVED.