About

Jonathan Cagan is the George Tallman and Florence Barrett Ladd Professor in Engineering at Carnegie Mellon University. His career spans collaborative and innovative work in education, research, and industry. Cagan researches engineering design automation and methods, merging AI, machine learning, and optimization methods with cognitive science problem solving. One focal area is the cognitive basis and computational modeling of designer processes to improve the effectiveness of human designers. Another area includes computational methods for the design and diagnosis of biomechanical systems. An additional focus is in user-centered design and integrated product development practice. In the CMU way, he has collaborated with designers, engineers, psychologists, neuro-scientists, marketers, computer scientists, and architects in his work. At Carnegie Mellon, Cagan co-founded and co-directed the Integrated Innovation Institute. He served as Associate Dean for Graduate and Faculty Affairs, Chief Academic Officer, and Interim Dean of the College of Engineering. Cagan was recently honored with the Robert A. Doherty Award for Sustained Contributions to Excellence in Education. Active in professional societies and editorial boards, Cagan is a Fellow in the American Society of Mechanical Engineers and was awarded with the ASME Design Theory and Methodology Award. He has authored several books, over 250 publications, and is an inventor on multiple patents.

Research topics

• Computer Science
• Artificial Intelligence
• Knowledge management
• Engineering
• Psychology
• Human–computer interaction
• Radiology
• Pathology
• Biology
• Engineering management
• Data science
• Cognitive psychology
• Obstetrics
• Bioinformatics
• Social psychology
• Medicine

Selected publications

• Complete Pareto Front Generation for Multi-Objective Optimization Using Graphics Processing Unit

  2026-01-01

  articleOpen accessSenior author
  PublisherDOI

• Fine-Scale Feature Encoding in Geometric Pre-training for Engineering Surrogate Modeling

  Journal of Mechanical Design · 2026-05-22

  article

  Abstract AI-driven surrogate modeling has emerged as an effective alternative to physics-based simulations for 3D design, analysis, and manufacturing. These models use data-driven techniques to predict physical quantities that traditionally require computationally expensive simulations. However, the scarcity of labeled CAD-to-simulation datasets has motivated the development of self-supervised and foundation models, in which geometric representation learning is performed offline and later adapted to downstream tasks using limited labeled data. While promising, existing approaches often struggle in applications that require accurate preservation of fine-scale geometric details. This work introduces a self-supervised geometric representation learning method designed to capture fine-scale geometric features from non-parametric 3D models. Unlike traditional end-to-end surrogate models, the proposed approach decouples geometric feature extraction from downstream physics prediction by learning a latent representation guided solely by geometric reconstruction losses. Key components include near-zero-level signed distance field sampling and a batch-adaptive attention-weighted loss function, which together enhance sensitivity to subtle yet physically influential geometric variations. The proposed method is validated through two case studies involving high-dimensional design parameter regression, achieving coefficients of determination exceeding 0.98, as well as structural mechanics tasks that demonstrate strong few-shot prediction performance for reaction forces and deformation fields. Comparisons with parametric surrogate models further illustrate our method's ability to bridge geometric and physics-based representations, providing an effective surrogate modeling solution in data-scarce settings.

  PublisherDOI

• GPU-Based Global Optimization for Engineering Design

  Journal of Computing and Information Science in Engineering · 2026-04-28

  articleSenior author

  Abstract Design optimization has important applications in many engineering fields, where the goal is to find the best design within the available means. In many practical applications, finding the best design can bring significant profit, quality, and performance advantages. However, popular design optimization methods often become trapped in local optima and fail to find the global optimum that corresponds to the best design, leading to inconsistent or incorrect assumptions and applications. In this paper, a novel global optimization method designed for GPU-based massively parallel computing is introduced to efficiently enclose the global optimum for continuous design optimization problems where the objective and constraint functions have analytic expressions. Using interval arithmetic, coupled with the computational power of GPU, the method iteratively rules out the regions in the design space where the global optimum cannot exist and leaves a finite set of regions where the global optimum must exist. Because of the rigor of interval arithmetic, the method is guaranteed to enclose the global optimum in the regions within a user-specified width tolerance for design optimization problems, even in the presence of rounding errors. The GPU-based global optimization method is validated through two case studies of the Ackley function and launch vehicle design. The results show that the method successfully encloses the global optimum that corresponds to the best design in each case study, while both a typical gradient-based method and a popular heuristic method become trapped in local optima that correspond to inferior designs.

  PublisherDOI

• Global optimization tailored for graphics processing units: Complete and rigorous search for large-scale nonlinear minimization

  PNAS Nexus · 2026-04-01

  articleOpen accessSenior author

  This paper introduces a numerical method to enclose the global minimum of a nonlinear function subject to simple bounds on the variables. Using interval analysis, coupled with the computational power and architecture of graphics processing units (GPUs), the method iteratively rules out the regions in the search domain where the global minimum cannot exist and leaves a finite set of regions where the global minimum must exist. For effectiveness, because of the rigor of interval analysis, the method is guaranteed to enclose the global minimum even in the presence of rounding errors. For efficiency, the method employs a novel GPU-based single program, single data parallel programming style to circumvent major GPU performance bottlenecks, and a variable cycling technique is also integrated into the method to reduce computational cost when minimizing large-scale nonlinear functions. The method is validated by minimizing 11 benchmark test functions with scalable dimensions, including the well-known Ackley function, Griewank function, Levy function, Rastrigin function, and Rosenbrock function. These benchmark test functions represent grand challenges of global optimization, and enclosing the guaranteed global minimum of these benchmark test functions with >80 dimensions has not been reported in the literature. Our method completely searches the feasible domain and successfully encloses the guaranteed global minimum of these 11 benchmark test functions with up to 10,000 dimensions using only one GPU in a reasonable computation time, far exceeding the reported results in the literature due to the unique method design and implementation based on GPU architecture.

  PublisherDOI

• Global optimization tailored for graphics processing units: Complete and rigorous search for large-scale nonlinear minimization

  PNAS Nexus · 2026-04-01

  articleOpen accessSenior author

  This paper introduces a numerical method to enclose the global minimum of a nonlinear function subject to simple bounds on the variables. Using interval analysis, coupled with the computational power and architecture of graphics processing units (GPUs), the method iteratively rules out the regions in the search domain where the global minimum cannot exist and leaves a finite set of regions where the global minimum must exist. For effectiveness, because of the rigor of interval analysis, the method is guaranteed to enclose the global minimum even in the presence of rounding errors. For efficiency, the method employs a novel GPU-based single program, single data parallel programming style to circumvent major GPU performance bottlenecks, and a variable cycling technique is also integrated into the method to reduce computational cost when minimizing large-scale nonlinear functions. The method is validated by minimizing 11 benchmark test functions with scalable dimensions, including the well-known Ackley function, Griewank function, Levy function, Rastrigin function, and Rosenbrock function. These benchmark test functions represent grand challenges of global optimization, and enclosing the guaranteed global minimum of these benchmark test functions with >80 dimensions has not been reported in the literature. Our method completely searches the feasible domain and successfully encloses the guaranteed global minimum of these 11 benchmark test functions with up to 10,000 dimensions using only one GPU in a reasonable computation time, far exceeding the reported results in the literature due to the unique method design and implementation based on GPU architecture.

  PublisherDOI

• VIRL: Volume-Informed Representation Learning towards few-shot manufacturability estimation

  Journal of Intelligent Manufacturing · 2025-02-18 · 1 citations

  articleOpen access

  Abstract Designing for manufacturing poses significant challenges in part due to the computation bottleneck of Computer-Aided Manufacturing (CAM) simulations. Although deep learning as an alternative offers fast inference, its performance is dependently bounded by the need for abundant training data. Representation learning, particularly through pre-training, offers promise for few-shot learning, aiding in manufacturability tasks where data can be limited. This work introduces VIRL, a Volume-Informed Representation Learning approach to pre-train a 3D geometric encoder. The pretrained model is evaluated across four manufacturability indicators obtained from CAM simulations: subtractive machining (SM) time, additive manufacturing (AM) time, residual von Mises stress, and blade collisions during Laser Power Bed Fusion process. Across all case studies, the model pre-trained by VIRL shows substantial enhancements in generalizability, as measured by R2 regression results, with improved performance on limited data and superior predictive accuracy with larger datasets. Regarding deployment strategy, case-specific phenomenon exists where finetuning VIRL-pretrained models adversely affects AM tasks with limited data but benefits SM time prediction. Moreover, the efficacy of Low-rank adaptation (LoRA), which balances between probing and finetuning, is explored. LoRA shows stable performance akin to probing with limited data, while achieving a higher upper bound than probing as data size increases, without the computational costs of finetuning. Furthermore, static normalization of manufacturing indicators consistently performs well across tasks, while dynamic normalization enhances performance when a reliable task dependent input is available.

  PublisherOA PDFDOI

• Guiding Generalized Team Problem-Solving Through a Collective Intelligence-Based Artificial Intelligence Facilitator

  Journal of Mechanical Design · 2025-06-11 · 6 citations

  article

  Abstract The use of artificial intelligence (AI) to guide team dynamics has the potential to transform collaborative problem-solving processes. Existing approaches to doing so are trained on prior problem-specific data, limiting them to problems that have already been solved. This research aims to extend AI-based approaches to novel situations by eliminating the need for prior data. This is accomplished by focusing on team communication and collective intelligence (CI) rather than problem-specific strategies. CI is a team's general ability to work well across various tasks and is more predictive of team performance than individual intelligence. This work introduces an AI facilitator that monitors CI attributes—collective attention, equal participation, and consistent communication—in real time and intervenes as necessary to guide teams toward better collaboration and overall performance. Two human subjects studies are performed on teams working together to design a mechanical system to test the AI facilitator. The studies vary in the structure of the problem-solving environment (virtual or colocated) and communication modality (text-only or verbal). The studies' findings support that the AI facilitator leads teams to better performance than without using the facilitator, and equivalent performance when compared to a human facilitator. This contribution to the field of AI in team management is notable because of the elimination of the need for prior data, making it applicable to novel situations. This work lays the groundwork for a new approach to potentially transform the future of teamwork.

  PublisherDOI

• CARA: The Corporate AI Readiness Assessment

  2025-08-17

  article

  Abstract Artificial Intelligence (AI) adoption is rapidly transforming industries, yet many organizations struggle to assess their readiness for AI integration. This paper introduces the Corporate AI Readiness Assessment (CARA), a tool designed to evaluate an organization’s preparedness for AI implementation across three critical dimensions: organizational readiness, workforce readiness, and technology readiness. CARA provides a structured framework that enables companies to identify strengths, gaps, and necessary actions to support AI-driven initiatives. While applicable across various industries, this assessment is particularly valuable for Small and Medium enterprises (SMEs) in manufacturing and engineering design, where resource constraints often present challenges in AI adoption. The assessment is structured as a survey, offering a scoring system that classifies organizations into four readiness categories: Emerging, Developing, Advancing, and Leading. By evaluating strategic alignment, workforce capabilities, and technological infrastructure, CARA facilitates informed decision-making for AI adoption. To demonstrate its applicability, this paper presents two case studies of organizations with different levels of AI maturity, highlighting how CARA’s insights can guide strategic planning. Future work will focus on validating CARA through broader industry applications and refining its scoring methodology. By providing a practical, scalable approach to AI readiness assessment, CARA aims to support organizations in navigating the complexities of digital transformation effectively.

  PublisherDOI

• Expanding the Generative Power of Large Language Models for Design Through Formal Design Grammars and Languages

  Journal of Computing and Information Science in Engineering · 2025-10-09 · 3 citations

  articleOpen access

  Abstract Research in design grammars has been underway for over 50 years and has demonstrated great generative power for a wide range of design and engineering domains. A key limitation, though, is the lack of support for designers to develop and computationally implement formal design grammars. We explore the potential of large language models (LLMs) to act as a collaborative grammar development partner that works with human designers and provides guidance during grammar development, as well as serving as a grammar interpreter that converts natural language descriptions of design grammars into executable python code. Methods for both interpreting previously known design grammars as well as interactively and collaboratively developing a new design grammar that is not known a priori are proposed. Three case studies, namely a truss design grammar, a half-hexagon shape grammar, and a technical process grammar, are investigated, covering string, shape, and graph grammars to explore the advantages and limitations of combining design grammars and LLMs. Finally, we position formal design grammars to be a key element for the future to expand the generative power of LLMs and enable them to become more repeatable, precise, and explainable for generative design tasks.

  PublisherOA PDFDOI

• Enclosing the Global Optimum: A GPU-Based Global Optimization Method for Engineering Design

  2025-08-17

  articleSenior author

  Abstract Design optimization is prevalent in many engineering fields, where the goal is to find the best design within the available means. Finding the best design can bring significant profit, quality, and performance advantages for engineering design and system engineering. However, popular design optimization methods, such as gradient-based methods and heuristic methods, often fail to find the best design and its corresponding global optimum, leading to inconsistent or incorrect assumptions and applications. In this paper, a novel GPU-based global optimization method is introduced to efficiently enclose the global optimum based on user-specified tolerances for design optimization problems. Using interval arithmetic, coupled with the computational power of GPU, the method iteratively rules out the suboptimal regions in the design space where the global optimum cannot exist and leaves a finite set of regions where the global optimum must exist. Because of the rigor of interval arithmetic, the method is guaranteed to enclose the global optimum for design optimization problems even in the presence of rounding errors. In addition, the method is originally designed for GPU-based massively parallel computing and therefore can efficiently utilize the computational power of the users’ GPU(s). The GPU-based global optimization method is validated through a three-stage launch vehicle design case study. The case study shows that the method can successfully enclose the guaranteed global optimum and find the best design in a reasonable computation time.

  PublisherDOI

Recent grants

• EAGER: Innovative Energy Farm Design

  NSF · $66k · 2009–2011

• Determining Consumer Preference Through an Interactive Virtual Reality Experience

  NSF · $375k · 2012–2016

• MOD: Stimulating Creative Insight - A Cohesive Model of Design Innovation Across Individuals, Groups and Computer Agents

  NSF · $214k · 2008–2011

• GOALI: Capturing, Implementing, and Generating Product Brand Through Shape Grammars

  NSF · $301k · 2003–2007

• SGER: Understanding the Role of Impasses and Representation Changes in Creative Design: An Initial Study

  NSF · $154k · 2006–2007

Frequent coauthors

• Kenneth Kotovsky

  Carnegie Mellon University

  106 shared

• Christopher McComb

  72 shared

• Philip R. LeDuc

  Université Grenoble Alpes

  27 shared

• Christian D. Schunn

  University of Pittsburgh

  26 shared

• Kosa Goucher-Lambert

  University of California, Berkeley

  25 shared

• Guanglu Zhang

  22 shared

• Nicolás F. Soria Zurita

  Pennsylvania State University

  21 shared

• Joshua T. Gyory

  Carnegie Mellon University

  19 shared

Awards & honors

• Robert A. Doherty Award for Sustained Contributions to Excel…
• Fellow in the American Society of Mechanical Engineers
• ASME Design Theory and Methodology Award