top of page


Anchor 1

I am a Postdoctoral Researcher at Meta, where I work on advancing the frontiers of artificial intelligence (AI) research and applying it to real-world problems. I have a PhD in Computer Engineering from Duke University, with a focus on computer vision, machine learning, and causal analysis. Now seeking research positions.






Amazon AWS

Quantization of Latent Diffusion Models

Meta Platforms Inc., Research Advisor: Hongbo Zhang

Developed an efficient quantization strategy for LDMs.
Increased efficiency of calibration using a single-step sampling.

Causal Understanding of Discriminative Self-supervised Learning

Duke University, Research Advisor: Yiran Chen

Developed a causal framework to understand discriminative SSL.
Explained undesired behaviors during inference.
Proposed solutions to overcome the undesired behaviors with solid theory support.

Federated Unsupervised Representation Learning

Duke University, Research Advisor: Yiran Chen

Implemented an unsupervised distributed learning framework. Proposed personalized network branches at local clients.

Unsupervised Image Segmentation

Samsung Semiconductor, Research Advisor: Mostafa El-Khamy

Implemented a weakly supervised loss in addition to the cluster loss. Designed a box sampling process for better label continuity in neighboring pixels.

Causal Framework for Generalized Image Representation

Duke University, Research Advisor: Taoyang Chen

Developed causal graphs for image representation learning. Synthesized observational data for validating causal graphs. Implemented experiments to test assumptions. Adapted causal relation to real images to improve the generalization of the representation.

Unsupervised Visual Representation Learning

ABB Inc., Research Advisor: Remus Boca

Representation Learning using a contrastive learning framework. Improved representations by inducing adversarial noise. Verified quality representations on vision downstream tasks.

Positive Unlabeled Learning in Object Detection


Duke University, Research Advisor: Kevin Liang

Verified the performance drop using a popular dataset that has incomplete labels. Improved performance using Positive Unlabeled learning instead of Positive Negative Learning is commonly used. Implemented PU detectors with Faster Regional CNN and Single Shot Detector.

Interpretability and Transfer Learning of Deep Neural Network



Duke University, Research Mentor: Prof. Henry Pfister

Trained deep CNN models with both synthesized and real image datasets and tested them on a real dataset. Utilized Class Activation Mapping to localize multiple objects in images and interpreted the inner mechanism of the network.


Multiple Instance Learning for Training Weakly Supervised Learning Models



Duke University, Research Mentor: Dr. Jordan Malof

Built MI SVM and CNN models to study the underlying relationships between the complexity of a classifier, the positive ratio in a bag, and the accuracy of the classifier.


Prediction of Seizure in EEG and ECG signals



Duke University, Research Mentors: Professor Leslie M. Collins and Dr. Nicholas Czernek

Built feature extractors and reduced dimension with PCA and GA. Trained a model with seizure and non-seizure segments with the Random Forest algorithm.



Maximizing Bit Rate in Cognitive Radio System



Research Mentors: Prof. Mohamed-Slim Alouini and Dr. Osama Amin

Verified simulations of different modulation schemes against their analytical expressions. Derived an analytical expression for error probability of improper signaling of secondary users.


C++/C       > 5 years

Matlab     > 5 years

Python     > 3 years

Java          > 1 year

PHP          > 1 year

MySQL     > 1 year

Duke University

Aug.2019 - May.2023


Electrical and Computer Engineering, advised by Professor Lawrence Carin and Professor Yiran Chen.


Duke University

Aug.2016 - May.2018


Electrical and Computer Engineering, Master of Science, GPA 3.82.

Master’s Project: “Multiple Instance Learning in Deep Learning” advised by Dr. Jordan Malof and Professor Leslie Collins. Courses focus on learning machine learning algorithms and applications in NLP, Computer Vision, and Pattern Recognition.


University of Cambridge

Oct.2014 - Jun.2015


Electrical and Electronics Engineering, Master of Engineering, GPA 3.70.

Master’s Thesis: “Design and build of an audio amplifier” advised by Professor Richard McMahon. Focuses on designing and testing chip hardware like VLSI.


University of Cambridge

Oct.2011 - Jun.2014


Electrical and Electronics Engineering, Bachelor of Arts, GPA 3.64.

Three years of bachelor's degree program focusing on general engineering aspects including mathematics, information, electrical and electronics, aerodynamics, mechanics, and bioengineering. 


Professional experience
Postdoctoral Researcher,
Meta Platforms Inc., Remote, USA



Researched on image generative AI models.
Quantized Stable Diffusion and deploy on mobile devices.
Developed research solutions to recommendation and generative applications.

Research Scientist Intern,
Samsung Semiconductor., Remote, USA



Researched unsupervised Image Segmentation.

Perception and Computer Vision Research Intern, ABB Inc., Raleigh, USA



Researched unsupervised representation learning to classify hyperspectral images. Researched multiple object tracking.

Software Engineer, Aqueti Intl., Durham, USA



Researched systems to maximize the storage efficiency in a camera system. Built scripts to determine the maximum writing rates for the camera memory system.

Software Engineer, Wavebot, Beijing, China



Tested different laser hardware, modified laser-based SLAM, and programmed code to optimize the SLAM performance of an indoor service robot.

Software Engineer, Cambridge Silicon Radio, Cambridge, UK


Designed the A4WP wireless charging transmitter including a beacon procedure and a time module in C and Python and performed dynamic and static parameters exchange between the receiver and the transmitter.

Y. Yang*, K. Liang*, and L. Carin, "Object Detection as a Positive Unlabeled Problem", The British Machine Vision Conference (BMVC), Aug 2020
D Wang*, Y Yang*, C Tao, F Kong, R Henao, L Carin, "Proactive Pseudo-Intervention: Causally Informed Contrastive Learning For Interpretable Vision Models", preprint
Y Chen, A Li, H Yang, T Zhang, Y Yang, H Li, S Banerjee, M Pajic, "AI-Powered IoT System at the Edge", International Conference on Cognitive Machine Intelligence (CogMi), Dec 2021
Guo Q, Chen J, Wang D, Yang Y, Deng X, Carin L, Li F, Tao C, "Tight mutual information estimation with contrastive Fenchel-Legendre optimization", Conference on Neural Information Processing Systems (NeurIPS), Nov 2022
Yang, Y., Li, H., & Chen, Y. "Stable and Causal Inference for Discriminative Self-supervised Deep Visual Representations". International Conference on Computer Vision (ICCV). Aug 2023

bottom of page