Photo of Mehdi Tarkian

Mehdi Tarkian

Senior Associate Professor

My research focuses on engineering automation through artificial intelligence and structured knowledge. I develop methods that enable access to engineering information, and perform complex engineering tasks in a traceable and reliable way.

Research

How can we reduce repetitive engineering work while allowing engineers to focus on decisions that require experience, creativity and judgement?

 I have worked with Design Automation since 2007. My earlier research focused on Knowledge-Based Engineering, parametric CAD, multidisciplinary optimisation and automated product development.

Recent advances in artificial intelligence have significantly expanded what can be automated. My current research therefore combines traditional engineering automation with large language models, agent-based systems, digital twins, computer vision, machine learning and structured knowledge representations.

I am particularly interested in systems where AI does not operate as an isolated black box. Instead, language models are combined with explicit data structures, engineering models, tools and validation mechanisms.

Teaching

My teaching primarily concerns product modelling, design automation and computational methods for product development. I am currently examiner for

TMKT57 – Product Modelling

TMKU01 – Design Automation of Customized Products

 

 

Publications

2026

Ali Kheirandish Koshkooi, Marie Jonsson, Mehdi Tarkian (2026) Spatio-Temporal Graph Neural Network Surrogate Modeling for Predicting Sheet Metal Deformation in a Novel Forming Process 12TH SWEDISH PRODUCTION SYMPOSIUM, 2026, Article 012058 (Conference paper) https://dx.doi.org/10.1088/1757-899X/1342/1/012058
Hamideh Pourrasoul Ouzi, Marie Jonsson, Mehdi Tarkian (2026) Agentic Framework for Production Time Estimation of Sheet Metal Parts from Engineering Drawings Using Multi-Modal Document Analysis 12TH SWEDISH PRODUCTION SYMPOSIUM, 2026, Article 012062 (Conference paper) https://dx.doi.org/10.1088/1757-899X/1342/1/012062
Sanjay Nambiar, Oscar Ikechukwu, Rahul Chiramel Paul, Marie Jonsson, Mehdi Tarkian (2026) Digital Twin-enabled adaptive robotics: multi-agent reasoning over language, vision, and structured database Production & Manufacturing Research, Vol. 14, Article 2708530 (Article in journal) https://dx.doi.org/10.1080/21693277.2026.2708530
Ali Kheirandish Koshkooi, Marie Jonsson, Mehdi Tarkian (2026) Time-dependent surrogate modeling for metal sheet deformation using random forests for a dieless sheet metal forming process Production Engineering, Vol. 20, Article 37 (Article in journal) https://dx.doi.org/10.1007/s11740-026-01426-6

2025

Sanjay Nambiar, Rahul Chiramel Paul, Oscar Chigozie Ikechukwu, Marie Jonsson, Mehdi Tarkian (2025) Digital Twin-Enabled Adaptive Robotics: Leveraging Large Language Models in Isaac Sim for Unstructured Environments Machines, Vol. 13, Article 620 (Article in journal) https://dx.doi.org/10.3390/machines13070620

Research

From Design Automation to Automation of Engineering Systems

Current research areas

My work currently centres around several closely connected areas:

Agentic engineering systems
Development of modular AI-agent architectures capable of retrieving information, using engineering tools, reasoning over structured data and executing multi-step engineering workflows.

Structured engineering knowledge
Methods for representing evidence, requirements, relationships, decisions and provenance explicitly rather than relying solely on information stored inside an LLM context. The objective is to make AI-supported engineering more traceable, auditable and reliable.

Digital twins and adaptive automation
Combining simulation, structured databases, sensors, computer vision and AI agents to create digital twins that can support increasingly autonomous industrial and robotic systems.

Engineering information retrieval and automation
Transforming information contained in engineering documents, drawings, standards, databases and other heterogeneous sources into accessible and machine-actionable knowledge. An important objective is to enable downstream automation in design, manufacturing, production preparation, verification and quality assurance.

AI-supported design and manufacturing
Application of machine learning, optimisation, physics-informed methods and generative approaches to CAD, product development, manufacturing and production systems.

A new car in the manufacturing area.

Automation Lab

Product development was revolutionized by the introduction of the first CAD software in 1963, a digital drawing board called Sketchpad. The premise of the innovation was to automate a manual, repetitive and erroneous draft-drawing procecess.

Co-workers

Organisation