KANANART KUWARANANCHAROEN
Machine Learning | Mathematical Optimization | Robotics
EDUCATION
PURDUE UNIVERSITY West Lafayette, Indiana, United States
Doctor of Philosophy August 2023
Major: Electrical and Computer Engineering, GPA 4.00/4.00
CHULALONGKORN UNIVERSITY Pathum Wan, Bangkok, Thailand
Bachelor of Engineering June 2016
Major: Electrical Engineering, GPA 3.90/4.00
First-class Honors, Gold Medal, Class Rank: 1
PROFESSIONAL EXPERIENCE
INTEL CORPORATION Hillsboro, Oregon, United States
AI Research Engineer/Scientist,
Strategic CAD & Heterogeneous Platform Lab, Intel Labs October 2023 – November 2024
Area of Research: Reinforcement Learning (RL), Binary Neural Network (BNN)
Technical Skills: Machine Learning (ML), Online RL, Offline RL, Safe RL, BNN
Software Skills: Python (PyTorch, JAX, Stable-Baselines3), MuJoCo simulator, Git, Linux
• Designed and implemented a novel machine learning technique for integrating Binary Neural Network
architecture within deep RL frameworks, developing efficient controllers for both software and hardware.
This innovation significantly improved system performance while reducing power consumption and
storage requirements, making the controllers more suitable for resource-constrained environments
• Applied advanced state augmentation techniques from the RL literature to ensure that online RL
algorithms comply with critical constraints while maintaining performance levels after training
• Expanded the application of offline RL algorithms to enhance power efficiency and performance in
software and hardware systems, successfully deploying a high-performing controller that leverages
existing data for safety-critical applications
• Utilized Git for version control, ensuring seamless collaboration across teams and maintaining
codebase integrity during the development
INTEL CORPORATION Hillsboro, Oregon, United States
Research Intern,
System Design and Architecture, Intel Labs September 2021 – April 2022
Area of Research: Reinforcement Learning (RL)
Technical Skills: Machine Learning (ML), Model-Free Reinforcement Learning, Robotics
Software Skills: Python (PyTorch), MuJoCo simulator, Git, Linux
• Developed advanced model-free deep reinforcement learning algorithms, specifically building upon the
Soft Actor-Critic (SAC), and tailored for high-dimensional discrete action spaces, making them
particularly effective for resource allocation problems
• Built a machine learning pipeline that fed observations into the developed RL controller and output
actions to the target system, ensuring smooth integration and real-time decision-making
• Implemented RL algorithms to control humanoid robots within the MuJoCo simulator, facilitating
simulation-based training, and evaluated the effectiveness of the algorithms through visualizing the
resulting control policies
• Initiated and led bi-weekly research reading sessions focused on the latest advancements in RL
literature, fostering a culture of continuous learning and promoting innovative idea generation among
team members
1
PURDUE UNIVERSITY West Lafayette, Indiana, United States
Graduate Research Assistant,
Electrical and Computer Engineering August 2017 – August 2023
Areas of Research: Resilient Distributed Optimization, Convex Analysis, Reinforcement Learning (RL)
Advised by: Shreyas Sundaram, Marie Gordon Professor of Electrical and Computer Engineering
Technical Skills: Mathematical Programming, Algorithm Design, Fault-Tolerant System, Decentralized
Architecture, Convex Analysis, Mathematical Analysis, Machine Learning (ML), Model-Based Reinforcement
Learning (MBRL)
Software Skills: MATLAB & Simulink, Python (NumPy), LaTeX
• Conducted in-depth characterization of fundamental properties and structures of key machine learning
problems, including the optimization of sums of convex functions, resulting in novel insights into the set
of potential solutions for these problems
• Proposed novel scalable resilient multi-agent optimization algorithms for multi-dimensional convex
functions, capable of mitigating the impact of adversarial agents without prior knowledge of their
identities, and theoretically analyzed convergence properties of the algorithms
• Proposed a unified algorithmic framework for resilient multi-agent optimization algorithms, and derived
novel results including linear convergence to a neighborhood of the solution and approximate
consensus for state-of-the-art algorithms within the framework
• Proposed a novel model-based reinforcement learning scheme which efficiently leverages structural
knowledge of the underlying system, and analyzed the upper bound on the sample complexity of the
proposed approach
CHULALONGKORN UNIVERSITY Pathum Wan, Bangkok, Thailand
Research Assistant,
Electrical Engineering June – December 2016
Area of Research: Physical System Modeling
Technical Skills: System Modeling, Electrical Component Design, System Performance Analysis
Software Skills: SystemVision, VHDL-AMS
• Researched SystemVision, a cloud-based tool for designing and modeling electro-mechanical systems
and analyzed the properties of components to optimize system performance for target applications
• Designed and developed electrical components using VHDL-AMS in SystemVision, creating valuable
tools for demonstrating electro-mechanical systems to instructors and undergraduate students
NARA INSTITUTE OF SCIENCE AND TECHNOLOGY (NAIST) Ikoma, Nara, Japan
Summer Research Intern,
Photonic Device Science Lab, Materials Science May – August 2015
Area of Research: Integrated Circuit (IC) Design
Technical Skills: Analog Circuit Design, Layout Design, CMOS Process Technology
Software Skills: CAD (Cadence Virtuoso)
• Collaborated with an Associate Professor to design digital and analog components for CMOS image
sensors using 0.18 μm CMOS process technology
• Designed schematic diagrams for analog components, such as pixels, column amplifiers, and output
buffers, by optimizing the bias voltage and size of NMOS and PMOS transistors to achieve linear
performance, and conducted simulations using CAD (Cadence Virtuoso) to verify the designs
• Designed layout of the integrated circuits (IC) based on the schematic diagrams, optimizing for compact
size while maintaining desired performance levels
2
PUBLICATIONS
Google Scholar: Google Scholar Profile
Mathematical Optimization
• K. Kuwaranancharoen, L. Xin and S. Sundaram, “Resilient Distributed Optimization,” Chapter 72 in
Encyclopedia of Systems and Control Engineering (ESCE), 2025, In Production
• K. Kuwaranancharoen and S. Sundaram, “On the Geometric Convergence of Byzantine-Resilient
Distributed Optimization Algorithms,” SIAM Journal on Optimization (SIOPT), vol. 35, pp. 210-239, 2025
[Paper Link]
• K. Kuwaranancharoen, L. Xin and S. Sundaram, “Scalable Distributed Optimization of Multi-
Dimensional Functions Despite Byzantine Adversaries,” in IEEE Transactions on Signal and Information
Processing over Networks (TSIPN), vol. 10, pp. 360-375, 2024 [Paper Link]
• K. Kuwaranancharoen, L. Xin and S. Sundaram, “Byzantine-Resilient Distributed Optimization of Multi-
Dimensional Functions,” in IEEE American Control Conference (ACC), Denver, CO, USA, 2020, pp.
4399-4404 [Paper Link]
Convex Analysis
• K. Kuwaranancharoen and S. Sundaram, “The Minimizer of the Sum of Two Strongly Convex
Functions,” Optimization, 2024, pp. 1-41 [Paper Link]
• K. Kuwaranancharoen and S. Sundaram, “On the Set of Possible Minimizers of a Sum of Known and
Unknown Functions,” in IEEE American Control Conference (ACC), Denver, CO, USA, 2020, pp. 106-
111 [Paper Link]
• K. Kuwaranancharoen and S. Sundaram, “On the Location of the Minimizer of the Sum of Two
Strongly Convex Functions,” in IEEE Conference on Decision and Control (CDC), Miami, FL, USA,
2018, pp. 1769-1774 [Paper Link]
Reinforcement Learning
• K. Kuwaranancharoen, “Towards Resource-Efficient Controllers: Training Binary Neural Network
Actors in Reinforcement Learning,” (Planned submission to the International Conference on Machine
Learning (ICML) 2025)
• J. Shen, K. Kuwaranancharoen, R. Ayoub, P. Mercati and S. Sundaram, “On the Benefits of
Leveraging Structural Information in Planning Over the Learned Model,” in IEEE American Control
Conference (ACC), San Diego, CA, USA, 2023, pp. 4555-4560 [Paper Link]
Miscellanies
• M. Hadisurya, L. Li, K. Kuwaranancharoen, X. Wu, Z. Lee, R. Alcalay, S. Padmanabhan, W. Tao and
A. Iliuk, “Quantitative Proteomics and Phosphoproteomics of Urinary Extracellular Vesicles Define
Putative Diagnostic Biosignatures for Parkinson’s Disease,” in Communications Medicine, 2023, 3(1),
pp. 64 [Paper Link]
3
NATURAL LANGUAGE PROCESSING PROJECTS
PROJECT: Named Entity Recognition with Fine-Tuned BERT on CoNLL-2003
Technical Skills: Natural Language Processing (NLP), Large Language Models (LLMs), Named Entity
Recognition (NER), Data Preprocessing, Fine-Tuning
Software Skills: Python (PyTorch, Hugging Face Transformers)
GitHub Repository: ner-bert-conll
• Developed an end-to-end pipeline for Named Entity Recognition (NER) using a BERT-based model
and the CoNLL-2003 dataset, covering data parsing, tokenization, and label alignment for token
classification
• Fine-tuned a pre-trained BERT model for NER tasks using Hugging Face's Trainer, achieving a high
F1 score in token classification, demonstrating the model's effectiveness in extracting named entities
• Created an intuitive inference function to extract named entities from text, facilitating easy deployment
and application of the trained model on new datasets
PROJECT: News Classification with Fine-Tuned GPT Model
Technical Skills: Large Language Models (LLMs), Fine-Tuning, Sematic Classification
Software Skills: Python (PyTorch, Transformers, PEFT)
GitHub Repository: LoRA-GPT2-AG-News-Classification
• Fine-tuned a GPT-2 model for classifying AG News articles, significantly enhancing classification
accuracy compared to the original pre-trained model
• Implemented a custom training loop to fine-tune the model, utilizing Low-Rank Adaptation (LoRA) for
the attention layers and adding a new classification head to optimize performance
• Conducted hyperparameter tuning for the LoRA rank to optimize performance, ensuring an efficient
balance between model complexity and memory usage while fine-tuning the GPT-2 model
PROJECT: Retrieval Augmented Generation (RAG) for Wikipedia Event Pages
Technical Skills: Large Language Models (LLMs), Retrieval Augmented Generation (RAG)
Software Skills: Python (OpenAI, tiktoken)
GitHub Repository: wiki-RAG
• Developed a RAG model to improve content generation by dynamically integrating relevant
information from Wikipedia event pages, enabling the model to better capture real-time context and
enhance relevance for queries on current events
• Produced dense embeddings by leveraging OpenAI’s high-performance text-embedding-3-small
model to capture nuanced content associations
• Integrated relevant data from this vector database into prompts for the GPT-4o model, facilitating
accurate and context-sensitive responses on recent Wikipedia events
COURSE PROJECT (Purdue University): Estimation of Damage and Recovery from Natural Disasters
Technical Skills: Machine Learning, Supervised Learning, Unsupervised Learning, Natural Language
Processing (NLP), Big Data Analytics, Data Visualization
Software Skills: Python (NumPy, pandas, Matplotlib, Scikit-learn)
Project Documentation: disaster-damage-estimation
• Designed and trained machine learning models for real-time estimation of damage caused by natural
disasters in impacted locations, utilizing Twitter data
o Applied Natural Language Processing (NLP) techniques for sentiment analysis on a massive
dataset of over 10 million tweets during Hurricane Sandy
o Utilized unsupervised learning techniques to cluster regions with similar damage levels and
provided dynamic maps for visualizing the predicted and official ground truth damage levels over
time, demonstrating high accuracy of the model
4
COMPUTER VISION PROJECTS
PROJECT: Denoising Diffusion Probabilistic Model (DDPM) from Scratch
Technical Skills: Computer Vision, Generative Models, Image Generation, U-Net Architecture
Software Skills: Python (PyTorch, torchvision)
GitHub Repository: ddpm-from-scratch
• Implemented a Denoising Diffusion Probabilistic Model (DDPM) from scratch using PyTorch, covering
the end-to-end pipeline from data processing to noise scheduling and reverse diffusion
• Designed and trained the model on the Stanford Cars dataset to generate high-quality images of cars,
using a U-Net architecture for reverse diffusion
• Developed modularized code for easy experimentation with different datasets and architectures,
including visualization tools for inspecting the diffusion process and evaluating model performance
PROJECT: Subject/Background Inpainting App
Technical Skills: Computer Vision, Image Segmentation, Diffusion Models, Segment Anything Model
(SAM), User Interface Design
Software Skills: Python (Gradio, PyTorch, Transformers, Diffusers)
GitHub Repository: subject-background-inpainting-app
• Developed an interactive web app that allows users to select a subject in an uploaded image by
clicking, using the Segment Anything Model (SAM) to automatically generate an accurate mask
• Integrated a text-to-image diffusion model to inpaint new backgrounds or subjects based on user-
provided natural language prompts, enabling creative image generation
• Enhanced user control by allowing the selection of classifier-free guidance scale, enabling a balanced
trade-off between adhering to the text prompt and fostering creative outputs
PROJECT: Aerial-to-Map Translation for Enhanced Navigation Using GAN
Technical Skills: Computer Vision, Generative Adversarial Network (GAN)
Software Skills: Python (PyTorch, torchvision)
• Trained a Pix2Pix GAN model specifically for satellite-to-map conversion, leveraging image-to-image
translation techniques to generate accurate and navigable map routes from aerial satellite imagery
• Implemented a U-Net generator with PatchGAN discriminator to enhance image realism and detail,
focusing on accurately capturing road structures, landmarks, and route connections in the generated
maps
PROJECT: Real-Time Object Detection for Autonomous Driving
Technical Skills: Computer Vision, Convolutional Neural Network (CNN), Object Detection
Software Skills: Python (TensorFlow, OpenCV, Keras)
• Implemented the YOLO (You Only Look Once) model for real-time object detection using images
captured from a car-mounted camera system, enabling the identification of various objects such as
pedestrians, vehicles, and traffic signs
• Optimized the YOLO model through hyperparameter tuning and data augmentation techniques,
resulting in improved accuracy and faster inference times on the target dataset, which comprised
diverse driving scenarios and environmental conditions
5
ROBOTICS AND CONTROL SYSTEMS PROJECTS
CAPSTONE PROJECT: Self-Driving Car Software Stack Development
Technical Skills: Robotics, Image Processing, State Estimation (Kalman Filter), Sensor Fusion, Finite State
Machines, Motion Planning, Controller Design
Software Skills: Python (NumPy, Scipy, OpenCV), CARLA simulator
GitHub Repository: self-driving-cars
• Developed a comprehensive software stack for self-driving cars within the CARLA simulator
environment, encompassing environmental perception, environmental maps construction, motion
planning, and vehicle controller
o Longitudinal and Lateral Control: Implemented advanced controllers to navigate the vehicle
along a predefined racetrack, adjusting speeds at specific waypoints to optimize performance
o Localization with ES-EKF: Utilized the Error-State Extended Kalman Filter (ES-EKF) to accurately
localize the vehicle. The filter seamlessly integrated IMU data for state prediction and incorporated
GPS and LIDAR position updates for precise corrections
o Visual Odometry: Developed a visual odometry system using a monocular camera mounted on
the vehicle, allowing for the estimation of the vehicle's trajectory based on image data
o Environmental Perception: Designed a robust environmental perception stack to extract crucial
scene information, ensuring self-driving cars could navigate their surroundings safely and reliably
o Motion Planning: Implemented a versatile motion planning system capable of dynamic obstacle
avoidance while adhering to the centerline of the lane and effectively responding to traffic signs
CAPSTONE PROJECT: Trajectory Planning and Control for a Mobile Manipulator
Technical Skills: Robotics, Motion Planning, Controller Design, State Estimation, Software Development
Software Skills: Python (NumPy), C++, CoppeliaSim, Robot Operating System (ROS)
GitHub Repository: mobile-manipulation
• Developed software to autonomously drive the KUKA youBot, a mobile base with four mecanum wheels
and a 5R robot arm, to manipulate a block from its initial to goal configuration in a simulation
environment using CoppeliaSim
o Designed and implemented a nominal trajectory planner for the robot’s end-effector
o Developed a feedback controller to mitigate effects from initial errors and disturbances, ensuring
precise execution of the planned trajectory
o Utilized odometry to accurately calculate the motion of the chassis based on the controller’s
commanded wheel velocities, optimizing overall system performance
SENIOR PROJECT (Chulalongkorn University): Backstepping Controller Design for PDEs
Technical Skills: Control Systems Design, Partial Differential Equations (PDEs), Numerical Methods
Software Skills: MATLAB & Simulink
• Designed feedback controllers for stabilizing linear partial differential equations (PDEs), including
systems such as reaction-advection-diffusion and quantum mechanical systems, utilizing the
backstepping method to ensure system stability
• Conducted in-depth literature research on boundary control design for nonlinear parabolic PDEs,
attempting to replicate and validate existing results
• Implemented numerical methods to simulate both uncontrolled and controlled dynamics of the target
PDE systems, exploring the impact of various parameter variations to evaluate system behavior and
controller effectiveness
6
SELECTED CERTIFICATES
Online Courses through Coursera
• Self-Driving Cars Specialization, University of Toronto, 2023
• Modern Robotics: Mechanics, Planning, and Control Specialization, Northwestern University, 2023
• Generative Adversarial Networks (GANs) Specialization, [Link], 2023
• Stochastic Processes, National Research University Higher School of Economics, 2019
• Deep Learning Specialization, [Link], 2018 (including computer vision and NLP)
• Python for Everybody Specialization, University of Michigan, 2018 (including SQL)
• Understanding Einstein: The Special Theory of Relativity, Stanford University, 2018
• Probabilistic Graphical Models Specialization, Stanford University, 2017
• Game Theory and Game Theory II: Advanced Applications, Stanford University & The University of
British Columbia, 2017
• Introduction to Thermodynamics: Transferring Energy from Here to There, University of Michigan, 2017
• Digital Signal Processing, École Polytechnique Fédérale de Lausanne, 2017
• Machine Learning, Stanford University, 2017
Additional Online Courses
• Quantum Mechanics: A First Course, MITx, 2021
• Paradox and Infinity, MITx, 2020 (intersection of philosophy and mathematics)
• Special Relativity: A Math-Based Introduction, World Science U, 2018 (with distinction)
SELECTED PAST ACTIVITIES
PURDUE UNIVERSITY
Soft Skills: Leadership, Strategic Planning, Organizational Skills, Interpersonal Skills, Communication Skills,
Problem-Solving Skills, Time Management
• Senior Consultant (2020-2021), Graduate Student Coordinator (2019) and Vice President (2018) of
Purdue University Thai Student Association
o Devised and implemented comprehensive long-term plans to enhance organizational efficiency,
resulting in significant improvements in productivity, cost savings, and overall performance
o Managed and executed cultural events, including a Sports Day Event for Thai students from
multiple universities and the Thai Food Festival, attended by Purdue students
CHULALONGKORN UNIVERSITY
Soft Skills: Leadership, Teamwork, Interpersonal Skills, Communication Skills, Event Planning and
Organization, Budget Planning, Initiative and Innovation
• Division Leader, College of Engineering, 2012-2016
o Led a division of engineering students, overseeing budget planning and event organization
o Represented the group at faculty meetings, providing updates on group activities and initiatives
o Provided supervision and support to younger group members in organizing and executing activities
• President of Academic Affairs, 2013-2015 Student Committee, Electrical Engineering Department
• Co-founder in the 1st Electrical Engineering Rural Development Camp, 2015
• President of the 12th Vishnukummabutra Camp (pre-engineering camp), 2014
• Academic Section Staff in the 6th FE Camp (pre-admission camp), 2013
7
ADDITIONAL INFORMATION
• Languages: Fluent in English, and native proficiency in Thai
• Additional Software Skills: MS Office, Java
• Awards and Achievements
o 1st Rank in Academics, 2012-2015, Electrical Engineering, Chulalongkorn University
o Earned 3rd place in the Research Poster Competition at the 21st Annual Information Security
Symposium hosted by the Center for Education and Research in Information Assurance and
Security (CERIAS)
o Selected as a finalist for the Center for Resilient Infrastructures, Systems, and Processes (CRISP)
Student Research Competition 2021-22
• Teaching Experience
o Provided part-time tutoring for mathematics and physics-related subjects, including undergraduate-
level courses in mathematics, signals and systems, automatic control, statistical learning, and deep
learning
